Ab undergraduate thesis_final
Upcoming SlideShare
Loading in...5
×
 

Ab undergraduate thesis_final

on

  • 399 views

 

Statistics

Views

Total Views
399
Views on SlideShare
385
Embed Views
14

Actions

Likes
0
Downloads
3
Comments
0

1 Embed 14

http://www.linkedin.com 14

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    Ab undergraduate thesis_final Ab undergraduate thesis_final Document Transcript

    • Impacts of Road Construction on Mangrove Structure in Atasta, Mexico Using GIS and Landsat TM Imagery A thesis submitted in partial fulfillment of the requirements for the honours degree of Bachelor of Science at Trent University Peterborough, Ontario Amber Brant April 2011 -0-
    • TABLE OF CONTENTSAbstract………………………………………………………………………………………………………………………………………………….3List of Tables…………………………………………………………………………………………………………………………………………..4List of Figures……………………………………………………………………………….………………………………….………………….…5Acknowledgements………………………………………………………………………………..………………………………….………….71.0 INTRODUCTION……………………………………………………………………………………………………………………8 1.1 Status, ecology and disturbance of mangroves……………………………………………………………….……8 1.2 Urban development in Campeche, Mexico…………………………………………………………………………14 1.3 Remote sensing in mangrove research ……………………………………….………………………….………....15 1.4 Research question…………………………….………………………….……………………………….……………………18 1.5 Research objectives……………….……………………………………….………………………………………………....19 1.6 Hypotheses and predictions……………………...…………………...………………………………………………...20 1.7 Approach………….……………………………………….……………………………………….……………………………….212.0 METHODS……………………………………………………………………………………………………………………….…..22 2.1 Study area………...………………………………………………………….………………………………………….…………22 2.2 Field sampling……….……………….............……………………………………………………………………………...27 2.3 Simpson’s biodiversity of mangroves………….…………………………………………………….……...………30 2.4 Remote sensing methods…………………...……………………………………………………………………….…...31 2.5 Distance-effect analysis……………….…….……………..……………………………………..………………....…..353.0 RESULTS…………………………………………….………………………………….…….......................................................37 3.1 Mangrove composition……………………………………………………………………………………………………….37 3.2 Selection of suitable SVI predictors…………………………………………………………………………………….40 3.3 Estimation and change in test variables……………………………………………………………………....…….42 3.3.1 Relative abundance of R. mangle………………………………………………………………………..42 3.3.2 Relative abundance of A. germinans…………………………………………………………………..47 3.3.3 Simpson’s biodiversity of mangroves………………………………………………………………….52 3.6 Distance-effect correlations………………………………………………………………………………………………..57 3.6.1 Change in R. mangle with distance from road……………………………………………..…..…57 3.6.2 Change in A. germinans with distance from road………………………………………………..60 3.6.3 Change in Simpson’s biodiversity with distance from road………………………..…….…634.0 DISCUSSION………………………………………………….………………………………………………………………………...66 4.1 Effect of road on R. mangle and A. germinans ………………………………………………………………..66 4.2 Effect of road on Simpson’s biodiversity………………………………………………………………..………..69 4.3 Efficacy of SVI in prediction of test variables …………………………………………………………………….70 4.4 Limitations of study……………………………………………………………………………….………………………….735.0 CONCLUSIONS AND RECOMMENDATIONS……………………………………..………………………………..….….74 -1-
    • 6.0 REFERENCES…………………..………………………………………………………………….………………………….………757.0 APPENDICES……………….………………………………………………………………………….….…………….…………….80 6.1 Appendix A: Raw field data…………….………………….……………………………………………..……..…80 6.2 Appendix B: ANOVA results in curve fitting for candidates for suitable SVI for predicting relative abundance of R. mangle…….……………………………………………….……81 6.3 Appendix C: ANOVA results in curve fitting for candidates for suitable SVI for predicting relative abundance of A. germinans.……………………………………………….……87 6.4 Appendix D: ANOVA results in curve fitting for candidates for suitable SVI for predicting Simpson’s biodiversity (1-D) of mangroves………….........………………….……92 -2-
    • Abstract A road was constructed through a mangrove forest in Atasta Lagoon, Mexico in 1986. Theimpacts of the road on mangrove structure were examined using field sampling, multispectral satelliteimage analysis and GIS applications. It was hypothesized that the construction of the road negativelyimpacts the: i) relative abundance of red mangrove (Rhizophora mangle); ii) relative abundance of blackmangrove (Avicennia germinans); and iii) biodiversity of all four species of mangrove. Sixteen 900m2 field quadrats were sampled in the lagoon for total abundance of each of the fourmangrove species: R. mangle, L. racemosa, A. germinans and C. erectus L. Simpson’s biodiversity index(1-D) was determined for each field plot. Regression analyses were used to select a suitable spectralvegetation index (SVI) for predicting relative abundance of R. mangle, of A. germinans and Simpson’sbiodiversity of the four species. NDWI, GEMI and EVI were selected for predicting relative abundance ofR. mangle (R2=0.34, p<0.05, df=15), A. germinans (R2=0.37, p=0.13, df=15) and Simpson’s biodiversityindex (R2=0.64, p<0.01, df=15), respectively. Using eight 650-m digital transects in ArcGIS, change from 1984-2009 for each of the threevariables were correlated with distance from road. There were significant correlations between distancefrom road and change in R. mangle in 3 of the 8 transects (p<0.05) and A. germinans in 2 of the 8transects (p<0.05). No significant correlations were found between distance and Simpson’s biodiversity.Results demonstrate that relative species abundance and Simpson’s biodiversity of mangroves can beeffectively predicted using SVI for large-scale change studies. Results suggest that the road negativelyimpacts the relative abundance of R. mangle. The impacts of the road on relative abundance of A.germinans and biodiversity of mangroves are inconclusive. -3-
    • List of TablesTable 1.1 Wavelengths and spatial resolution of the 7 bands detected by Landsat 5 TM ........................10Table 2.1 Center coordinates of 30x30m quadrats for inventory of mangroves…….............................……20Table 2.2 Selected spectral vegetation indices (SVI) used as candidates to predict the three test variables……………………………………………………………………………………………………………………….…25Table 3.1 Statistics for estimates and change of relative abundance of R. mangle in Atasta Lagoon for 1984 and 2009.......................................................................................................…37Table 3.2 Statistics for estimation and change in relative abundance of A. germinans in Atasta Lagoon for 1984 and 2009.……..............................................................................................……42Table 3.3 Statistics for estimated values and change in Simpson’s biodiversity (1-D) in Atasta Lagoon for 1984 and 2009......................................................................................................…47 -4-
    • List of FiguresFigure 2.1 Location of study area in southeast Mexico within the APFFLT………………………………………….…15Figure 2.2 Lagoons found within the APFFLT ……………………………....................................................…………15Figure 2.3 Display of Landsat 5 TM Band 5 (1.55-1.75 µm) for 24-year progression of change before and after the construction of a road in the Atasta Lagoon..………………………………………………..…16Figure 2.4 Regeneration on abandoned road crossing the mangrove forest.………………………….…………..…17Figure 2.5 Locations of the 30x30m field plot for inventory of mangroves.……………………………………………19Figure 2.6 Display of pixels with 30-m resolution for Landsat TM Band 5.……………………………………..………20Figure 2.7 Locations of digital transects in ArcGIS.……………….……………………………………………………..…………27Figure 3.1 Total mangrove abundance counts and corresponding Simpson’s biodiversity index values (1-D) for 30x30m quadrats in Atasta Lagoon…………………….………………………….………..…29Figure 3.2 Predictive relationship between NDWI and relative abundance of R. mangle.………………....…32Figure 3.3 Predictive relationship between GEMI and relative abundance of A. germinans…………………32Figure 3.4 Predictive relationship between EVI and Simpson’s biodiversity index (1-D)………………….……32Figure 3.5 Estimated relative abundance of R. mangle in Atasta Lagoon for 1984 and 2009 using NDWI-based arithmetic operations.….……………………………………………………………………….……...…34Figure 3.6 Estimated change in relative abundance of R. mangle in Atasta Lagoon using NDWI-based arithmetic operations..……………………………………………….……………………………………35Figure 3.7 Frequency distribution of estimated values for relative abundance of R. mangle in Atasta Lagoon for 1984 and 2009.………………………………………………………………………………..………36Figure 3.8 Frequency distribution of estimated values for change in relative abundance of R. mangle in Atasta Lagoon from 1984-2009.……………….………………………………………………………36Figure 3.9 Estimated relative abundance of A. geminans in Atasta Lagoon for 1984 and 2009 using GEMI-based arithmetic operations.….……………………………………………………………………….……...…39Figure 3.10 Estimated change in relative abundance of A. geminans in Atasta Lagoon using GEMI-based arithmetic operations..……………………………………………….……………………………………40Figure 3.11 Frequency distribution of estimated values for relative abundance of A. geminans in Atasta Lagoon for 1984 and 2009.………………………………………………………………………………..………41 -5-
    • Figure 3.12 Frequency distribution of estimated values for change in relative abundance of A. geminans in Atasta Lagoon from 1984-2009.……………….………………………………………..…………41Figure 3.13 Estimated Simpson’s biodiversity (1-D) of mangroves in Atasta Lagoon for 1984 and 2009 using EVI-based arithmetic operations.….…………………………………………………………….…….…...…44Figure 3.14 Estimated change in Simpson’s biodiversity (1-D) of mangroves in Atasta Lagoon using NDWI-based arithmetic operations..……………………………………………….……………………………………45Figure 3.15 Frequency distribution of estimated values for Simpson’s biodiversity (1-D) of mangroves in Atasta Lagoon for 1984 and 2009.…………………………………………………………………46Figure 3.16 Frequency distribution of estimated values for change in Simpson’s biodiversity (1-D) of mangroves in Atasta Lagoon from 1984-2009.……………….……………………………………….………46Figure 3.17 Regression curves for correlation between distance from road and change in relative abundance of R. mangle in Atasta Lagoon from 1984 to 2009……………………………………..………49Figure 3.18 Regression curves for correlation between distance from road and change in relative abundance of A. germinans in Atasta Lagoon from 1984 to 2009…………………………………..…..52Figure 3.19 Regression curves for correlation between distance from road and change in biodiversity of mangroves in Atasta Lagoon from 1984 to 2009…………………………………….……55 -6-
    • Acknowledgements First and foremost, I would like to thank my supervisor of the Department of Geography, Dr.Raul Ponce-Hernandez. His encouragement in the imagination and organization of this project is nowinvaluable to me. This being my first international research project, Dr. Ponce allowed me to overcomeany borders that presented itself in the logistics of solidifying a study like this. I would also like to giveDr. Ponce-Hernandez my full gratitude for all of the experiences that has been made available to bringthis project to life, including the trip with Trent University’s Integrated Watershed Management coursein May 2010 to Campeche, Mexico, where many of the ideas presented here were first developed.Finally, I would like to thank Dr. Ponce-Hernandez for the countless conversations in epistemology; I ama better philosophy student because of these exchanges. My gratitude also goes to Dr. David Beresford, my supervisor in the Department of Biology.While Dr. Ponce provided much technical and educational support, Dr. Beresford was also invaluable inthe clarification of ideas that surfaced during the course of this project. Dr. Angel Sol-Sanchez, of the Colegio Postgraduados Tabasco was also a key player in thisproject. His field assistance and logistical support throughout my stay in Mexico will be alwaysappreciated. Also, Mario Dominiguez of the Colegio Postgraduados Tabasco was a huge support in thefield, and his help is very much appreciated throughout this process. I would like to give Estrella Perez (M.Sc. candidate, Baja California) my fullest gratitude for hergenerous help with the field aspect of this project. Her assistance was invaluable, and my Spanish hasimproved because of her. Also, thanks must be given to Asuncion, the owner of the fishing boat inwhich we used to travel throughout the lagoon. Edgar Tomes of the geospatial department for theColegio Postgraduados Tabasco was also a huge help in the GIS support for this project, as well as a hugehelp in field support. -7-
    • 1.0 INTRODUCTION1.1 Status, ecology and disturbance in mangroves Mangroves are coastal forests that are found in estuaries, along riverbanks and in shallowlagoons in 124 countries worldwide (FAO, 2007). They are the only forest systems that can inhabit theharsh buffer zone between terrestrial and ocean environments (Alongi, 2002). There is speculationsurrounding the evolution of mangroves, but a popular opinion is that their origins are terrestrial treesin the Indo-west Pacific just after the arrival of the angiosperms 114 million years ago, where extendedperiods of wetness allowed for a transition from dry to brackish adaptations in these plants (Kathiresanand Bingham, 2001). Mangrove fossils are also found in areas where they no longer exist, like Texas, USAand Western Australia, demonstrating that mangroves have persisted through many paleoclimaticevents and associated changes (Kathiresan and Bingham, 2001). Mangroves have been studied extensively over the past 40 years, mostly under the subject ofproductivity and community ecology, but the recent worldwide extent of mangrove forests has beenunder review and has been inaccurately represented in the past, due to lack of inclusiveness and reliabledata (FAO, 2007). Giri et al. 2010 used high quality Landsat, Quickbird and Ikonos satellite images toestimate the worldwide extent of mangroves in 2000 at 13.8 million ha using data from 118 countries,the most comprehensive and reliable estimate made to date. There are approximately 741,917 ha ofmangrove forests in Mexico, which accounts for 5.4% of the global mangrove coverage (Giri et al. 2010).The coast along the Gulf of Mexico is more humid than that of the Pacific coast, and the high diversityand abundance of mangrove forests along the gulf reflects this difference in humidity (López-Portillo andEzcurra, 2002). Mangrove forests found in the states of Veracruz, Tabasco and Campeche typically havea greater average height and species richness of mangroves, as the temperature in these states rarely -8-
    • falls below 14°C (López-Portillo and Ezcurra, 2002). In the state of Campeche, the extent of mangroves isestimated at 196,552 ha as of 2009 (CONABIO, 2009), which is 26% of the country’s total extent, ascalculated from Giri et al. 2010. Coastal ecosystems in the tropics are undergoing an increasing amount of change. Increasingpopulation pressure and demands for economic stability in tropical countries have placed theseecosystems in a vulnerable state. Based on the limitations for mangrove persistence worldwide, thefuture of this taxonomic group is influenced largely by climate change and by anthropogenicmanipulations (Alongi, 2002). Mangroves have many ecosystem services that are economically and socially valuable. Theyshelter floods from entering the inland areas along coasts and when the forests are large in stemdiameter, survival rates from tsunamis are high (Yanagisawa et al. 2009). With a large stem size,inundation of water inland from a tsunami in Thailand was reduced by 30% when the tsunami depth wasless than 3m high (Yanagisawa et al. 2009). There are six types of mangrove forest, based on water inputs and topography: riverine (river-based), fringe (ocean-based), basin (interior), overwash (island), hammock and scrub, all of whichprovide different ecosystem services (Lugo and Snedaker, 1974; Ewel et al. 1998). In terms ofmaintaining ecosystem integrity, mangroves trap sediments, process nutrients from freshwater systemsand provide essential habitat for many wildlife species (Ewel et al. 1998). Riverine forests are mostimportant for conservation in terms of sediment trapping, because they are in the closest proximity tofreshwater systems and in their absence, sedimentation of particles can lead to erosion and offshoredeposition of these particles (Ewel et al. 1998). The above-ground biomass of basin mangrove forests ishigher than any other aquatic ecosystem, rivalling even the densest rainforests (Alongi, 2002). Fringeforests protect shorelines and provide food and habitat for wildlife (Ewel et al. 1998). -9-
    • Mangrove-fishery linkages may be the most economically-biased justification for theirconservation. The state of Campeche, Mexico produces one sixth of Mexico’s entire total shrimp outputand the shrimp fishery sector employs 13% of the employed population in the state (Barbier, 2000). Themain nurseries for these shrimp are the mangrove forests located around the Términos Lagoon, Mexicoand with an estimated 2km2 loss of these forests per year, costs associated with decreases in shrimpharvesting are estimated at $150,000USD per year (Barbier, 2000). The value of mangroves as units of conservation has not been easy to assess in the past, due tolack of complete knowledge about their function and how they adapt to change over time. In mostcases, conservation is location- and economy-specific. For example, in 1995 the CINVESTAV-IPN UnidadMerida and the EPOMEX Program of the Universidad Autonoma de Campeche outlined four mainservices provided by mangroves of the state of Campeche: use as timber resource for housing andcharcoal production (estimated at $451USD/ha/year for charcoal and $631USD/ha/year for housing),fishery provisions (estimated at $1578 USD/ha/year), water filtering services (estimated at$1193USD/ha/year) and habitat for critical wildlife (Cabrera et al. 1998). Mangroves are a unique taxonomic group both structurally and functionally. Adaptations andattributes include: aerial prop roots, salt/water/carbon regulation, tide-dispersed propagules, viviparousembryonic reproduction and rapid canopy growth (Kethiresean and Bingham, 2001; Alongi, 2002). Manyfactors affect how mangroves are distributed at different spatial scales. In the global scale, mangrovesare limited by temperature and humidity (can only occupy areas between 30° S and 30°N latitudes)while at the regional scale, by rainfall and tidal frequency (Kathiresan and Bingham, 2001; Alongi, 2002).At a local scale, because they require inflow of nutrients from freshwater sources and well-circulatedwater flows, they are typically found in well-drained alluvial soils in well-sheltered areas (Kathiresan andBingham, 2001). - 10 -
    • Like most forest communities, mangroves are organized in distributional patterns related tospecies type. In these forests, species richness is very low; the understory layer is filled with seedlings ofthe overstory species, but the functional understory of herbaceous and shrub species does not exist(Alongi, 2002; Krauss et al. 2008). They are found in different levels of abundance and growth rates,depending on a suite of environmental conditions which include: frequency of floods, salinity, level ofand soil anoxia (Krauss et al. 2008). Mean leaf area size in the red mangrove (Rhizophora mangle) in Mexico is positively correlatedwith annual precipitation and latitude (Kathiresan and Bingham, 2001). Following the clearcut of a redmangrove forest on the north coast of Para, Brazil, Berger et al. 2006 demonstrated that after years ofsuccession involving different non-mangrove species, white mangrove and then black mangrove wereable to enter the area and establish themselves successfully, but even after approximately 10 years, thered mangrove cannot enter the area that it previously inhabited. López-Portillo and Ezcurra (1989) conducted a study in the Mecoacán Lagoon, just west ofFrontera in the state of Tabasco, Mexico, investigating the effect of salinity on height and diameter ofthe red (R. mangle), white (Laguncularia racemosa) and black mangrove (Avicennia germinans) found inthe area. Surrounding the lagoon, there are basins that are low in salinity and mudflats that are high insalinity (López-Portillo and Ezcurra, 1989). A. germinans was found in higher relative abundance in themudflats, while all three species were found in evenly distributed abundances in the basin areas (López-Portillo and Ezcurra, 1989). R. mangle is restricted to intertidal zones and lagoons of humid tropical countries (Dominguez etal. 1998). Variations observed in R. mangle due to habitat specialization include: tree morphology,dominance in ecosystem structure, leaf area size and fruit size (Dominguez et al. 1998). Flowers areobserved all year round in this species and seedlings typically establish close to the parent tree(Dominguez et al. 1998). Depending on resource availability, R. mangle can alter its leaf morphology, - 11 -
    • photosynthetic rates, plant stature and uptake of nutrients in an area (Farnsworth and Ellison, 1996).Under high canopy enclosure, this species can slow its growth rate in shade, and take advantage of a gapin the canopy, transitioning to high growth rates (Farnsworth and Ellison, 1993). Globally, the main issues surrounding mangrove conservation include: clearing for urbanexpansion and tourism, species introductions (R. mangle in Florida), road construction, agriculturalconversion, oil pollution, erosion, storm damage, conversion for aquaculture and use of herbicides(Farnsworth and Ellison, 1997). Disturbance is natural process that occurs in forest ecosystems, necessary for function andpromotion of species composition and succession that follows as a consequence (Sousa, 1984). Thediscrimination between disturbance and stress is important to establish in terms of mangrove functionand change. Natural disturbances observed in mangrove ecosystems include: hurricanes, lighting strikes,tidal fluctuations, extreme flood events (Sherman et al. 2000). Mangroves are considered very resilientin the face of natural disturbances due to the following adaptations: storage of reservoir nutrientsbelow-ground, high biotic turnover due to nutrient fluxes, internal re-use of resources like water andnutrients, rapid reconstruction post-disturbance, high abundance of keystone species associated withmangrove ecosystems and finally, positive and negative feedbacks that allow flexibility (Alongi, 2008). Despite their resilience, however, mangroves are as susceptible as any other forest ecosystemto stress-inducing disturbance. Ellison and Farnsworth (1996) outline four classes of anthropogenicdisturbance that have been observed for mangrove ecosystems. Firstly, large-scale extraction for woodproducts and fishery provisions alter soil pH and disrupt food-web linkages, respectively (Ellison andFarnsworth, 1996). Second, large-scale pollution events from petroleum, metals and sewage results inmassive defoliation followed by tree death at all biological stages of life (Ellison and Farnsworth, 1996).Third, reclamation in the form of land use change, including agriculture, urban development, tourismand aquaculture, disrupts mangrove ecosystems through deforestation or alteration of the forests - 12 -
    • (Ellison and Farnsworth, 1996). Lastly, climate change impacts such as elevated CO2 levels, sea-level rise,temperature increase and high-frequency storm events all contribute to changes in mangroveecosystems, effects that need further research to accurately address the issues (Ellison and Farnsworth,1996). - 13 -
    • 1.2 Urban development in Campeche, Mexico Until 1976, the main activities observed in the region of the Términos Lagoon were forestry,agriculture and small-scale fishery practices (Bach et al. 2005). Mexico has been involved with oilextraction, refinement and exportation for approximately 40 years, following the 1971 discovery ofCantarell, the major hydrocarbon deposit in the marine platform of the Términos Lagoon (Soto-Galera etal. 2010). Oil production began in the Sound of Campeche about five years after the discovery ofCantarell and now produces 80% of crude oil and 30% of natural gas for all of Mexico (Soto-Galera et al.2010). The rise of the oil industry in the state of Campeche has also influenced land use changes due toincreased urbanization and Petróleos Mexicanos (PEMEX) oil industry infrastructure (Soto-Galera et al.2010). From 1974-2001, the two main causes for land change surrounding the Términos Lagoon were i)increased urbanization and consequently agricultural land and ii) oil infrastructure establishment inplace of wetlands including mangroves (Soto-Galera, 2010). Mangrove forests decreased in extent in theTérminos Lagoon region by 13% from 1974-2001 (Soto-Galera et al. 2010). Oil and gas supplies in thisregion will last approximately two more decades, at which point the state of Campeche will enter a neweconomic state (Bach et al. 2005). Urban development in the Téminos Lagoon area has had social and environmental impactsincluding: extreme poverty and marginalization in the village of Atasta as an indirect result of PEMEXinfluence, oil pipeline leaks, water pollution from agricultural runoff and sewage, land modification foragriculture and cattle raising, construction of bridges between the island of Carmen and the mainlandthat disrupt wetland functioning, illegal fishing, mangrove deforestation for timber and roadconstruction that restrict water flow between ecosystems (Bach et al. 2005). - 14 -
    • 1.3 Remote sensing in mangrove research There is a growing popularity in the use of remotely sensed data from satellites in large-scaleecology studies (Aplin, 2005). The three main divisions of remote sensing in ecology are landclassification, ecosystem models using field measurements and land change detection (Aplin, 2005). Thebasis of orbital remote sensing is the collection of information from platforms which then collectelectromagnetic energy from the Earth’s surface, which is then transmitted, recorded and separatedinto bands with different wavelengths that can be then analyzed using geospatial software (Table 1.1)(Campbell, 2007). The software displays pixels with corresponding digital numbers that representspectral signatures gathered from the ground (Campbell, 2007). For vegetative classification and long-term vegetative change, especially at the species level, the sensors, Landsat Thematic Mapper (TM)and Landsat Enhanced Thematic Mapper (ETM+) have advantages over the other sensors in studyinglong-term changes in landscapes (Xie et al. 2008). Spatial resolution for a given sensor describes theminimum distance between two objects on the ground that can be discriminated (Campbell, 2007). Allmultispectral bands for Landsat TM data have 30-m resolution, and the thermal infrared band with 120-m resolution. For species-level discrimination, the two best-suited sensors are the IKONOS andQuickbird, with spatial resolutions of 4m and 2.4-2.8m for multispectral bands, respectively, but scenesproduced from these products are expensive and do not provide extensive historical data records (Xieet al. 2008). Vegetative mapping at the species level in heterogenous environments is challenging withLandsat satellite imagery, but can be done with adequate field data calibration (Xie et al. 2008).Remotely-sensed Landsat TM data have been used extensively in mangrove research. Mangrovemapping research deals with the accurate determination of the extent of mangroves in an area,facilitated by remote sensing methods (Long and Skewes, 1996). Landsat TM data have been used to - 15 -
    • classify mangroves (Long and Skewes, 1996; Ramírez-García et al. 1998; Sulong et al. 2002; Mas, 2004;D’iorio et al. 2007; Lee and Yeh, 2009) and determine change in extent over time (Kovacs et al. 2001;Béland et al. 2006). In these cases, the 30-m resolution of Landsat TM is suitable for the large-scaleapplications that are involved. The leaf area index (LAI) of mangrove trees can be measured in the field, and has beeneffectively correlated with band ratios of wavelength bands provided by Landsat TM data (Díaz andBlackburn, 2003). Spectral vegetative indices (SVI) are band ratios that have been developed, tested andused for their effectiveness in detecting reflected radiant energy from the ground and used forpredicting physical properties on the ground (Myeni et al. 1995; Baugh and Groeneveld, 2006).Produced from multiple bands, SVI are arithmetic expressions that can be computed from thewavelengths bands produced by satellite sensors (Campbell, 2007). Band 4 (near-infrared) from theLandsat TM sensor (expressed as TM4) has been used to represent vegetative density, greenness andphotosynthetic activity, as plants reflect energy at this wavelength (Mironga, 2004). TM3 can be used toexpress leaf area, as it is related to the plant’s absorption of the Sun’s energy (Mironga, 2004). TM5(shortwave-infrared) and TM7 (mid-infrared) represent wavelengths that are related to moisturecontent on the ground, and can be used to estimate biomass of plants and canopy closure, as it detectsreflected energy from the soil as well as the soil as well as the canopy (Mironga, 2004). The Simple Ratioindex (TM4/TM3) and the Normalized Difference Vegetation Index (NDVI) (TM4-TM3/TM4+TM3) aretwo examples of indices that have been tested for their reliability to estimate vegetative parameters(Baugh and Groeneveld, 2006). - 16 -
    • Table 1.1 Wavelengths and spatial resolution of the 7 bands detected by Landsat 5 TM (Campbell, 2007). Landsat 5 (TM sensor) Wavelength (µm) Resolution (meters) Band 1 (Blue) 0.45 - 0.52 30 Band 2 (Green) 0.52 - 0.60 30 Band 3 (Red) 0.63 - 0.69 30 Band 4 (Near Infared) 0.76 - 0.90 30 Band 5 (Shortwave Infrared 1.55 - 1.75 30 Band 6 (Thermal) 10.40 - 12.50 120 Band 7 (Mid Infrared) 2.08 - 2.35 30 - 17 -
    • 1.4 Research question An unpaved road was constructed in 1986 in the Atasta Peninsula, located south of the village ofAtasta, Campeche, Mexico. The road was constructed as a means for transportation of materials fromthe north to the south end of the area, but is no longer in operation. The elevated soil platform isapproximately 5m above the remainder of the forest, to prevent inundation onto the road. Chemicalslike asphalt and concrete were not used in construction or maintenance and traffic was infrequentduring operation. The study site provides an exceptional opportunity to study a known disturbance in anotherwise protected natural area. This study is the result of the research question, ‘how does theconstructed road impact surrounding mangrove structure in the area?’ - 18 -
    • 1.5 Research objectives The overall goal of this study is to effectively examine the direct impacts of a constructed roadon surrounding mangrove structure in Atasta Lagoon, Campeche, Mexico using field sampling,multispectral satellite image analysis and GIS applications. Change analysis as well as distance-effectanalysis will be the main methods to fully examine the effect of the constructed road. The specific aimsof the study are to: i) effectively predict canopy species composition and biodiversity of mangrovesusing spectral vegetation indices (SVI) and ii) evaluate the effect of the road on change in speciescomposition of two dominant species as well as their biodiversity in the area. - 19 -
    • 1.6 Hypotheses and predictions The basis for this research is to gain understanding in how a constructed road impacts mangrovestructure, in terms of species composition and biodiversity. Three hypotheses are proposed to supportthe research question and its underlying analysis. It is hereby proposed that the constructed roadnegatively impacts the i) relative abundance of red mangrove (Rhizophora mangle); ii) relativeabundance of black mangrove (Avicennia germinans); and iii) biodiversity of all species of mangrove inthe forest along a gradient from the location of the road. All hypotheses are mutually exclusive and onlysupported if canopy characteristics are effectively predicted using spectral vegetation indices (SVI). - 20 -
    • 1.7 Approach Due to the size of the area and theme of the overall research project, remote sensing methodswere used to address the research question. Ground measurements of mangrove community structurewere integrated with multispectral satellite image analysis and GIS applications. This integration relieson the statistical relationship between these ground measurements and spatial patterns detected bymultispectral satellite imagery, as well as manipulations of arithmetic formulas using multispectralbands given in the image data. - 21 -
    • 2.0 METHODS2.1 Study area The study area lies in the village of Atasta in the state of Campeche, Mexico between 18°34’07and 18°37’21 latitudes and -92°04’15 and -91°58’02 longitudes in the Área de Protección de Flora yFauna Laguna de Términos (APFFLT) (Figure 2.1). The Términos Lagoon, Atasta Lagoon, De CarlosLagoon, Pom Lagoon and Puerto Rico Lagoon are located within the APFFLT (Figure 2.2). Atasta Lagoonhas an area of 30 km2 with an average depth of 1.5 m, depending on the season (Ruiz-Marín et al. 2009).The sediment type of the lagoon is muddy-clay and experiences the dry season from February to May,rainy season from June to September and influence from north-east winds from October to January(Ruiz-Marín et al. 2009). Temperatures range from 25-31°C in the area, depending on the season (Ruiz-Marín et al. 2009). The APFFLT hosts the largest density of mangrove species in the state (Vega, 2005). The canopyvegetation is nearly 100% mangrove, with four species found within the study area: red (Rhizophoramangle), white (Laguncularia racemosa), black (Avicennia germinans) and the less common buttonmangrove (Conocarpus erectus L.). The red mangroves in the area are very common and typicallydominate the area, with black mangrove as a secondary species. The maximum heights reached by theR. mangle and A. germinans are approximately 20m and 30m, respectively. The average approximateage of these mangroves in the area is 20 years, with a few nearly 200 years old (personalcommunication). Laguncularia racemosa and Conocarpus erectus L. are less common and reachapproximate maximum heights of 18m and 10m, respectively. Palm, banana and cacti are also commonin certain areas. Biomass and basal area of mangrove forests increases with increasing distancewestward from the Términos lagoon, as humidity increases (Barreiro-Güemes, 1999). The Atasta lagoonis the second lagoon to the east, with little water renewal and dominated by large but sparse black - 22 -
    • mangroves (Barreiro-Güemes, 1999). The De Carlos lagoon is a shallow area with red mangroves aspioneers and black mangroves as the dominant species (Barreiro-Güemes, 1999). The road is located a southwest direction from Federal Highway 180 (Figure 2.3). The road wasconstructed as a means for transportation of materials from the north to the south end of the area, butis no longer in operation. The elevated soil platform is approximately 5m above the remainder of theforest, to prevent inundation onto the road. Chemicals like asphalt and concrete were not used inconstruction or maintenance and traffic was infrequent during operation. Regeneration has taken placeon the abandoned road including banana (Musa sp.), Citrus sp., and noni (Morinda citrifolia), a plantwhose fruits have medicinal qualities (Figure 2.4). The region received legal protection on June 6, 1994 through administrative efforts by the“Comisión Nacional de Áreas Naturales Protegidas” (CONANP) (Vega, 2005). A small village with apopulation of 2,096 as of 2005, Atasta is geographically located in close proximity to increasing urbandevelopment and oil extraction activities and infrastructure (Implan, 2010). The current issues in Atastainclude: oil exploration and infrastructure leading to human affection and disease, decreasing fishproductivity, loss of critical habitats due to deforestation, low agricultural productivity and sulphurdioxide emissions by PEMEX recompression stations in the area (Yáñez-Arancibia et al. 1999; Ruiz-Marínet al. 2009). A gasline rupture in 1985 in the area of the recompression station caused an increase insalinity in the area and consequently dry deposition of sulphur dioxide (Yáñez-Arancibia et al. 1999).Social issues also are apparent in the area, with conflicts occurring between PEMEX and the Movementof Fishers and Farmers of the Atasta Peninsula who have blockaded federal highway 180 demandingcompensation through public works (Bach et al. 2005). - 23 -
    • Figure 2.1 Location of study area in southeast Mexico within the APFFLT. Figure 2.2 Lagoons found within the APFFLT. - 24 -
    • 1986-01-15 1999-01-19 1986-07-26 2009-11-30 1987-04-24 2010-02-02Figure 2.3 Display of Landsat 5 TM Band 5 (1.55-1.75 µm) for 24-year progression of change before and after the construction of a road in the Atasta Lagoon. - 25 -
    • Figure 2.4 Regeneration at south end of the abandoned road crossing the mangrove forest. - 26 -
    • 2.2 Field sampling In order to examine the relationship between the road and its surroundings, an inventory of themangrove forest was made during the dry season, January 4-January 9th, using a boat to access thechannels between the mangrove forests. For the area, 17 field sampling points were selected prior todata collection to calibrate with the information given in satellite imagery (Table 2.1, Figure 2.5). Thepoints were selected with a judgemental bias – clusters of three were identified throughout the area tosample different habitats in order to maintain a high degree of heterogeneity in sampling. At each field point, a 30 x 30 m quadrat was assembled, with center coordinates pre-determinedand used for navigation, corresponding to one pixel displayed in satellite imagery (Figure 2.6). Ahandheld Garmin GPS receiver (GPSMAP 76CSx) was used to locate the field points with an averageposition error of ±3.96 m. A compass was used to align the edges of the quadrats in north-south aspects. Mangrove species were identified based on bark appearance, root structure and leafappearance. Identification was assisted by an experienced naturalist familiar with mangrove ecosystemsin the area. Abundance counts of adult R. mangle, L. racemosa, A. germinans, and C. erectus L. weredetermined and recorded within each quadrat using four 15 x 15m sections within each plot. Onlymangroves with a height of at least 2m were included in the inventory. Due to the fact that mangrovesoccupy nearly 100% of the overstory and understory canopy, other vegetation was noted but notassessed. Field quadrat 5 was inaccessible by foot due to a high density of mangrove prop roots andconsequently omitted from further analysis. - 27 -
    • Figure 2.5 Locations of the 30x30m field plot for inventory of mangroves. - 28 -
    • Table 2.1 Center coordinates of 30x30m quadrats for inventory of mangroves.Quadrat Easting Northing Longitude Latitude (m) (m) (degrees, minutes, seconds) (degrees, minutes, seconds) 1 599280 2059050 -92°03’31.9337 18°37’12.2438 2 599310 2058870 -92°03’30.9422 18°37’06.3828 3 599430 2058630 -92°03’26.8905 18°36’58.5545 4 601530 2057550 -92°02’15.4317 18°36’23.0569 5 601740 2057580 -92°02’08.2610 18°36’23.9962 6 601410 2056950 -92°02’19.6357 18°36’03.5583 7 599880 2056170 -92°03’11.9781 18°35’38.4473 8 599790 2056050 -92°03’15.0702 18°35’34.5588 9 599910 2055840 -92°03’11.0139 18°35’27.7064 10 601050 2054340 -92°02’32.3930 18°34’38.7110 11 601170 2054850 -92°02’28.2065 18°34’55.2818 12 601620 2054490 -92°02’12.9205 18°34’43.4919 13 602040 2055630 -92°01’58.3832 18°35’20.5055 14 602400 2055570 -92°01’46.1123 18°35’18.4905 15 602578 2055510 -91°01’40.0506 18°35’16.5072 16 603270 2058480 -92°01’15.8900 18°36’53.0059 17 603480 2058570 -92°01’08.7077 18°36’55.8965 Figure 2.6 Display of pixels with 30-m resolution for Landsat TM Band 5. - 29 -
    • 2.3 Simpson’s biodiversity of mangroves Simpson’s biodiversity index (1-D) (SBI) was used to estimate biodiversity for each field plot. Allfour species of mangrove were included in analysis of biodiversity. The index is geared towardsabundance of the dominant species, and therefore considered an indicator of dominance concentration(Hill, 1973). This index is valuable in this case where the two dominant mangrove species are found inhigh densities, with the other two species as secondary species. The equation for Simpson’s index is: SBI =1 - where ni = number of individuals in species i, n = total number of individuals Values approaching 1 suggest high biodiversity and values approaching 0 suggest lowbiodiversity (Hill, 1973). - 30 -
    • 2.4 Remote sensing methods2.4.1 Satellite data A Landsat TM satellite image acquired on November 30, 2009 was downloaded from the USGSGlobal Visualization Viewer (http://glovis.usgs.gov/) for the location with path/row, 21/47 (with centerof swath at 18.8° latitude and -91.4° longitude) and used as the most recent cloud-free image thatcorresponds with season for field collection. A second image with the same path and row was acquiredfor the date November 25, 1984. The 1984 image was used along with the 2009 image for changeanalysis, with the road construction occurring in 1986.2.4.2 Pre-processing The .tiff files were extracted and opened in PCI Geomatica© software for processing. The studyarea was clipped from all bands within the image bundle with the following extents: 18°34’06 to18°37’21 latitudes and -92°04’15 to -91°58’02 longitudes. Bands 1-5 and 7 were used for the study, allhaving equal spatial resolution. A high-pass edge sharpening filter was passed on all bands with a 33x33kernel size. This filtering process ensures that all pixel value possibilities (1-255) were used in digitalnumber representation, increasing variability in spectral signatures. - 31 -
    • 2.4.3 Processing of SVI The selection of 16 SVI candidates originated from primary literature from which reliableestimates of forest canopy characteristics have been demonstrated using calibration with fieldmeasurements (Table 2.2). For each SVI, a map was produced in PCI Geomatica© using the band ratioswithin the 2009-11-30 satellite image. Determination of values for each SVI was made using the rastercalculator within PCI’s interface for appropriate bands.2.4.4 SVI Predictions for test variables The 16 spectral vegetation indices (SVI) were tested for their strength in predicting each of thethree test variables: relative abundance of R. mangle, relative abundance of A. germinans and Simpson’sbiodiversity index (1-D). The SVI values produced using the 2009-11-30 image were considered in theanalysis. The 16 field-collected values for each of these were used as dependent variables in regressioncurve fitting for linear, logarithmic, inverse, quadratic, cubic, compound, power, S, growth, exponentialand logistic relationships using SPSS Statistics 17.0 software. The suitable SVI for each test variable wasselected based on goodness-of-fit with the field data. Those with a high coefficient of determination (R2)and low p-value based on ANOVA tests were used a qualifying candidates. The resulting equations produced from regression analyses were computed in the PCIGeomatica© raster calculator with the selected SVI as the independent variable to produce a mapdisplaying each of the three test variables. The same equation used for the 2009-11-30 satellite imagebundle was used also for the 1984-11-25 image bundle to produce maps displaying estimates of thethree test variables for this date. Finally, three change maps were produced displaying change in relative - 32 -
    • abundance of R. mangle, change in relative abundance of A. germinans and change in Simpson’sbiodiversity (1-D) from 1984-2009. In total, nine maps were produced: six displaying estimates of eachof the three test variables in 1984 and 2009 and three displaying change in the three test variables from1984 to 2009. The nine raster maps were saved as .pix files and opened in ArcGIS for further analysis. The .pixfiles were exported into raster format and clipped to the 2,660ha extent: 18°34’07 to 18°37’21 latitudesand -92°02’50 to -92°00’18 longitudes for display purposes in proximity to the road. - 33 -
    • Table 2.2 Selected spectral vegetation indices (SVI) used as candidates to predict the three test variables. SVI Formula for computation using digital numbers (DN) of pixels in both images TM4 Landsat 5 TM Band 4 TM5 Landsat 5 TM Band 5 TM7 Landsat 5 TM Band 7Normalized Difference Vegetation Index (NDVI) ([TM4-TM3] / [TM4+TM3]) (Gould, 2000) Structure Insensitive Pigment Index (SIPI) ([TM4-TM1] / [TM4-TM3]) (Sims and Gamon, 2002) Simple Ratio (SR) (TM4 / TM3) (Myeni et al. 1995) Normalized Difference Water Index (NDWI) ([TM4-TM5] / [TM4+TM5]) (Gao, 1996) Chlorophyll Vegetation Index (CVI) ([TM4/TM2] * [TM3/TM2]) (Vincini et al. 2008) Green Vegetation Index (GVI) ([TM4+TM5] / [TM3+TM7]) (Todd and Hoffer, 1998) Mid Infrared Index (MIRI) (TM5 / TM7) (Feeley et al. 2005) Difference Vegetation Indices (DVI) (TM4-TM3) (Diaz and Blackburn, 2003) (TM4-TM2) (TM3-TM2)Global Environmental Monitoring Index (GEMI) n(1 – (0.25*n)) – [(TM3 - 0.125) / (1 - TM3)] (Pinty and Verstraete, 1992) where n= [ 2*(TM42 – TM32) + (1.5*TM4)+(0.5*TM3)] / (TM3 + TM4 + 0.5) Enhanced Vegetation Index (EVI) 2.5* [ (TM4 – TM3) / (Huete et al. 1997) [TM4 + (6*TM3) – (7.5*TM1) + 1] ] Atmospherically Resistant Vegetation Index [TM4 – (2*TM3 – TM1)] / (ARVI) [TM4 + (2*TM3 – TM1)] (Kaufman and Tanré, 1996) - 34 -
    • 2.6 Distance-effect analyses In order to examine the effect of the constructed road on mangrove structure, 8 digital transectswere produced within ArcGIS as polyline shapefiles (Figure 2.7). The transects were positioned at 1000-m intervals perpendicular to the road, with a length of 650 m each. Values were obtained for pixelsfound every 30 m along the transect for the following variables: change in relative abundance of R.mangle from 1984-2009, change in relative abundance of A. germinans from 1984-2009 and change inSimpson’s biodiversity (1-D) from 1984-2009. In total, 22 values were found for each transect for each ofthe three test variables. The values found at each point along the transects were correlated with distance from road,with each test variable as the dependent variable and distance as the independent variable. Linearregression analyses were made for each transect with the three variables, with two divisions. Division 1includes initial distance from road to a visible saturation point, thereafter the effect of the road appearsto stabilize in division 2. Significant correlations were those considered with a high coefficient ofdetermination (R2) and low corresponding p-value. - 35 -
    • Figure 2.7 Locations of digital transects in ArcGIS. - 36 -
    • 3.0 RESULTS3.1 Mangrove composition Within the 16 field plots sampled, the four species of mangrove were found in various relativeabundances, the highest mangrove density located near the south of the lagoon (Figure 3.1a). The mostdominant mangrove species encountered was R. mangle with 966 individuals, followed by A. germinanswith 806 individuals. Secondary mangrove understory species included L. racemosa with 522 individualsand C. erectus L. with 39 individuals. There is a significant positive linear relationship between abundance of A. germinans andabundance of R. mangle over the sixteen 900m2 field quadrats sampled (R2=0.274, p<0.05, df=15),suggesting a lack of competition between these species in these areas (Figure 3.1b). There is nosignificant relationship between abundance of R. mangle and Simpson’s biodiversity (R2=0.1267,p=0.176, df=15) or between abundance of A. germinans and Simpson’s biodiversity (R2=0.0687, p=0.327,df-15) (Figure 3.1c-d). - 37 -
    • 300 0.8 275 0.7 250 225 0.6 Simpsons biodiversity (1-D) 200 0.5 175 Abundance 150 0.4 125 0.3 100 75 0.2 50 0.1 25 0 0 1 2 3 4 6 7 8 9 10 11 12 13 14 15 16 17 Quadrat #Figure 3.1a Total mangrove abundance counts (C. erectus L. , L. racemosa , A. germinans ,R. mangle ) and corresponding Simpson’s biodiversity index values (1-D ) for 30x30m quadratsin Atasta Lagoon. - 38 -
    • 200 y = 0.8x + 19 R² = 0.274 150 p<0.05, df=15R. mangle 100 50 0 0 20 40 60 80 100 A. germinans Figure 3.1b Relationship between total abundance of 2 A. germinans and R. mangle per 900m field quadrats. 1 y = 0.002x + 0.5 R² = 0.1267 0.8 p=0.176, df=15Simpsons (1-D) 0.6 0.4 0.2 0 0 20 40 60 80 100 120 140 R. mangle Figure 3.1c Relationship between total abundance of R. mangle and 2 Simpson’s biodiversity index (1-D) per 900m field quadrats. 1 y = 0.002x + 0.5 R² = 0.0687 0.8 p=0.327, df=15 Simpsons (1-D) 0.6 0.4 0.2 0 0 20 40 60 80 100 A. germinans Figure 3.1d Relationship between total abundance of A. germinans and 2 Simpson’s biodiversity index (1-D) per 900m field quadrats. - 39 -
    • 3.2 Selection of Suitable SVI Predictors For the prediction of relative abundance of R. mangle, NDWI was selected with a stronglogarithmic relationship (R2=0.3402, p=0.018, df=15), significant with 95% confidence and root-mean-square (RMS) value of 0.0096 (Figure 3.2). In predicting the relative abundance of A. germinans, GEMI was selected with a cubicrelationship (R2=0.3682, p=0.13, df=15), which has no statistical significance, but the goodness-of-fit andRMS value of 0.0101 permit this SVI as the most suitable candidate (Figure 3.3). In the case of A.germinans, MIRI demonstrates a higher predictive strength with a cubic relationship (R2=0.3811, p=0.11,df=15), but includes several outliers that influence the curve in a visibly biased shift. For the prediction of Simpson’s biodiversity index (1-D), the suitable SVI selected is EVI for itsstrong quadratic relationship (R2=0.6394, p=0.0013, df=15), significant with 99% confidence and an RMSof 0.0078 (Figure 3.4). - 40 -
    • 1.0 1.0 y = 0.85145234 ln(x) + 1.12787657 RMS=0.00965 R2 = 0.3402 Relative abundance of R. mangle 0.8 p=0.018 0.8 Predicted 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0 0.2 0.4 0.6 0.8 1 NDWI Observed Figure 3.2 Predictive relationship between NDWI and relative abundance of R. mangle. 1.0 1 y = 0.0000000000294x3 + 0.0000003027889x2 RMS=0.0101 Relative abundance of A. germinans + 0.0009694613309x + 1.3246147011149 0.8 R2 = 0.3682 0.8 p=0.13 0.6 0.6 Predicted 0.4 0.4 0.2 0.2 0.0 0 -8000 -6000 -4000 -2000 0 0 0.2 0.4 0.6 0.8 1 GEMI Observed Figure 3.3 Predictive relationship between GEMI and relative abundance of A. germinans. 1.0 1 y = -6.01044356x2 + 26.21671375x - 27.88633915 R2 = 0.6394 RMS=0.0078 0.8 p=0.0013 0.8 0.6 0.6 Predicted1-D 0.4 0.4 0.2 0.2 0.0 0 1.5 1.75 2 2.25 2.5 2.75 3 0 0.2 0.4 0.6 0.8 1 EVI Observed Figure 3.4 Predictive relationship between EVI and Simpson’s biodiversity index (1-D). - 41 -
    • 3.3 Estimation and change in test variables3.3.1 Relative abundance of R. mangle The estimated relative abundance of R. mangle is given for the years 1984 and 2009 in Figure3.5. It was expected that there would be a theme of negative change in relative abundance of R. manglefor the 29,500 pixels computed. There are three general clusters of land where drastic change occurs,just west of the middle section of the road, southeast of the road and along the periphery of the roaditself (Figure 3.6). The estimates for relative abundance of R. mangle in November 1984 and November 2009follow symmetrical unimodal frequency distributions across the raster map (Figure 3.7). In November1984, the estimated mean relative abundance of R. mangle for the 2,660 km2 area was 0.276 ± 0.201per 900m2 section of land for the area, the equivalent of one pixel computed (Table 3.1). The estimatedmean relative abundance decreased to 0.255 ± 0.199 per 900m2 section of land in November 2009(Table 3.1). The overall estimated change in relative abundance of R. mangle from November 1984 toNovember 2009 follows a symmetrical unimodal frequency distribution (Figure 3.8). The estimatedmean change in relative abundance from 1984 to 2009 is -0.021 ± 0.186 per 900m2 section of land,which is a 7.6% decrease from the estimated 1984 value (Table 3.1). - 42 -
    • Figure 3.5 Estimated relative abundance of R. mangle in Atasta Lagoon for 1984 and 2009 using NDWI-basedarithmetic operations. - 43 -
    • Figure 3.6 Estimated change in relative abundance of R. mangle in Atasta Lagoon using NDWI-based arithmeticoperations. - 44 -
    • Figure 3.7 Frequency distribution of estimated values for relative abundance of R. mangle in Atasta Lagoon for1984 and 2009. N=29500 pixels. Figure 3.8 Frequency distribution of estimated values for change in relative abundance of R. mangle in AtastaLagoon from 1984-2009. N=29500 pixels. - 45 -
    • Table 3.1 Statistics for estimates and change of relative abundance of R. mangle in Atasta Lagoon for 1984 and2009. N=29500 pixels. Year Minimum Maximum Mode Median Mean Standard Deviation 1984 0 1.000 0 0.324 0.276 0.201 2009 0 0.997 0 0.295 0.255 0.199 1984-2009 -1.000 1.000 0 0 -0.021 0.186 - 46 -
    • 3.3.2 Relative abundance of A. germinans It was expected that there would be positive change in relative abundance of A. germinans forthe 29,500 pixels computed. The estimated relative abundance of R. mangle is given for the years 1984and 2009 in Figure 3.9. There are two general regions where observable change occurs: just west of themiddle section of the road and southeast of the road, in the same areas where there is decrease inrelative abundance of R. mangle (Figure 3.6; Figure 3.10). There is no visible change along the peripheryof the road for this species (Figure 3.10). For both November 1984 and November 2009, the relative abundance estimates of A.germinans follow right-skewed frequency distributions across the raster map (Figure 3.11). In November1984, the estimated mean relative abundance of A. germinans for the 2,660ha area is 0.301 ± 0.204 per900m2 section of land for the area (Table 3.2). The estimated mean relative abundance increased to0.315 ± 0.222 per 900m2 section of land in November 2009 (Table 3.2). The overall estimated change in relative abundance of A. germinans from November 1984 toNovember 2009 follows a symmetrical trimodal distribution (Figure 3.12). The estimated mean changein relative abundance from 1984 to 2009 is 0.014 ± 0.243 per 900m2 section of land, which is a 4.6%decrease from estimated 1984 value (Table 3.2). - 47 -
    • Figure 3.9 Estimation of relative abundance of A. germinans in Atasta Lagoon for 1984 and 2009 using GEMI-basedarithmetic operations. - 48 -
    • Figure 3.10 Estimated change in relative abundance of A. germinans from 1984 to 2009 in Atasta Lagoon usingGEMI-based arithmetic operations. - 49 -
    • Figure 3.11 Frequency distribution of estimated values for relative abundance of A. germinans in Atasta Lagoon for1984 and 2009. N=29500 pixels. Figure 3.12 Frequency distribution of estimated values for change in relative abundance of A. germinans in AtastaLagoon from 1984-2009. N=29500 pixels. - 50 -
    • Table 3.2 Statistics for estimation and change in relative abundance of A. germinans in Atasta Lagoon for 1984 and2009. N=29500 pixels. Year Minimum Maximum Mode Median Mean Standard Deviation 1984 0 0.977 0 0.348 0.301 0.204 2009 0 0.997 0 0.351 0.315 0.222 1984-2009 -0.977 0.977 0 0 0.014 0.243 - 51 -
    • 3.3.3 Simpson’s biodiversity of mangroves The estimated relative abundance of R. mangle is given for the years 1984 and 2009 in Figure3.13. It was expected that there would be an overall negative change in Simpson’s biodiversity for the29,500 pixels computed. There are three areas of land where visible change occurs: just west of themiddle section of the road and southeast of the road, in the same areas where there is change inrelative abundance of the two dominant species, and finally along the north periphery of the road(Figure 3.14). For both November 1984 and November 2009, the Simpson’s biodiversity estimates followleft-skewed frequency distributions across the raster map (Figure 3.15). In November 1984, theestimated mean Simpson’s biodiversity for the 2,660ha area is 0.471 ± 0.282 per 900m2 section of land(Table 3.3). The estimated Simpson’s biodiversity decreased to 0.441 ± 0.287 per 900m2 section of landin November 2009 (Table 3.3). The overall estimated change in Simpson’s biodiversity from November 1984 to November 2009follows an asymmetrical trimodal distribution (Figure 3.16). The estimated mean change from 1984 to2009 is -0.030 ± 0.242 per 900m2 section of land, which is a 6.4% decrease from the estimated 1984value (Table 3.3). - 52 -
    • Figure 3.13 Estimated Simpson’s biodiversity (1-D) of mangroves in Atasta Lagoon from 1984 to 2009 using EVI-based arithmetic operations. - 53 -
    • Figure 3.14 Estimated change in estimation of Simpson’s biodiversity (1-D) in Atasta Lagoon using EVI-basedarithmetic operations. - 54 -
    • Figure 3.15 Histogram for distribution of estimated values for Simpson’s biodiversity (1-D) in Atasta Lagoon for1984 and 2009. N=29500 pixels.Figure 3.16 Histogram for distribution of estimated values for change in Simpson’s biodiversity (1-D) of mangrovesin Atasta Lagoon from 1984-2009. N=29500 pixels. - 55 -
    • Table 3.3 Statistics for estimated values and change in Simpson’s biodiversity (1-D) in Atasta Lagoon for 1984 and2009. N=29500 pixels. Year Minimum Maximum Mode Median Mean Standard Deviation 1984 0 0.702 0 0.627 0.471 0.282 2009 0 0.702 0 0.597 0.441 0.287 1984-2009 -0.702 0.702 0 0 -0.030 0.242 - 56 -
    • 3.6 Distance-effect correlations The overall effect of the road on each of the test variables were tested using ANOVA for linearcorrelation. The overall effect was divided between a distance of 0-225m and 255-645m, based onobservations that saturation occurs at 255m, a distance after which the effect is predicted to bestabilized. For points where it is believed that water is encountered, in areas where mangroves are notfound, the corresponding data points were omitted from analysis and the next point used in the trend.Transects 7 and 8 are such examples, where distances of 105 and 135 m on transect 7 display water anddistances of 615 and 645 m on transect 8 display water. For all three test variables, the overall effect ofthe road is considered to have an impact if at least two of the eight total transects for a given distancedivision demonstrate a statistically and visibly significant correlation.3.6.1 Change in R. mangle with distance from road There are statistically and visibly significant positive linear correlations between distance fromthe road and change in relative abundance of R. mangle between 1984 and 2009 observed in the 0-225m division for transect 1 (R2=0.5691, p<0.05, df=7), 255-645m division for transect 4 (R2=0.3121,p<0.05, df=13) and 0-255m division for transect 5 (R2=0.5462, p<0.05, df=7) (Figure 3.17a-b). Fortransects 1-5, there are generally positive relationships between distance and change in abundance, upto the threshold of 225m (Figure 3.17a-b). After the 225m point, the effect appears to be stabilizing(Figure 3.17a-b). There is no evidence of any effect occurring in transect 6 for either division (Figure3.17b). Finally, for transects 7 and 8, there appears to be a great deal of noise in the data, likely due tohabitat variability and water conditions that can be observed in Figure 2.7 (Figure 3.17b). - 57 -
    • Transect 1 y = 0.0022x - 0.4105 y = 0.0004x - 0.1097 R² = 0.5691 R² = 0.2344 p<0.05, df=7 Change in R. mangle 0.4 p=0.079 0.2 -1E-15 -0.2 0 100 200 300 400 500 600 700 -0.4 -0.6 Distance (m) y = 0.0013x - 0.1446 Transect 2 y = -0.0003x + 0.0582 R² = 0.3841 R² = 0.0548 p=0.101 Change in R. mangle 0.4 p=0.421 0.2 -1E-15 -0.2 0 100 200 300 400 500 600 700 -0.4 -0.6 Distance (m) Transect 3 y = -0.0005x + 0.3093 y = 8E-05x - 0.0095 R² = 0.0025 R² = 0.1236 0.4 p=0.218 Change in R. mangle p=0.907 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m) Transect 4 y = 0.0014x - 0.3042 y = 0.0006x - 0.3019 R² = 0.456 R² = 0.3121 0.4 Change in R. mangle p=0.066 p<0.05, df=13 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m)Figure 3.17a Regression curves for transects 1-4 for correlation between distance from road and change in relative abundance of R. mangle in Atasta Lagoon from 1984 to 2009. Shaded are significant correlations. - 58 -
    • y = 0.0021x - 0.2424 Transect 5 y = -0.0002x + 0.1657 R² = 0.5462 R² = 0.0682 Change in R. mangle 0.4 p<0.05, df=7 p=0.197 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m) Transect 6 y = -0.0001x + 0.1387 y = -0.0005x + 0.1631 R² = 0.0382 R² = 0.1897 0.4 p=0.503 Change in R. mangle p=0.281 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m) y = -5E-05x + 0.0402 Transect 7 y = 0.0007x - 0.5458 R² = 0.0009 R² = 0.137 0.4 p=0.943 p=0.236 Change in R. mangle 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m) y = -0.0008x + 0.0579 y = 0.0006x - 0.5926 R² = 0.3098 Transect 8 R² = 0.2701 0.4 p=0.251 p=0.057 Change in R. mangle 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m)Figure 3.17b Regression curves for transects 5-8 for correlation between distance from road and change in relative abundance of R. mangle in Atasta Lagoon from 1984 to 2009. Shaded are significant correlations. - 59 -
    • 3.6.2 Change in A. germinans with distance from road There are statistically and visibly significant positive linear correlations between distance fromroad and change in relative abundance of A. germinans observed in division 255-645m for transect 1(R2=0.3262, p<0.05, df=13) and in division 255-645m for transect 2 (R2=0.3648, p<0.05, df=13)(Figure 3.18a). There are no statistically or visibly significant relationships observed in the remaining 6transects (Figure 3.18a-b). In transects 1 and 8, there is a slight negative correlation between distancefrom road and change in relative abundance for the 0-225m division, but only up to the 225m threshold(Figure 3.18a-b). As observed with change in relative abundance of R. mangle, transect 6 demonstratesno effect of the road, and transects 7 and 8 include a great deal of noise in the data (Figure 3.18b). - 60 -
    • y = -0.0005x + 0.0585 y = 0.0001x - 0.0547 0.3 Transect 1 R² = 0.2835 R² = 0.3262 Change in. A. germinans 0.2 p=0.174 p<0.05, df=13 0.1 0 -0.1 0 100 200 300 400 500 600 700 -0.2 -0.3 Distance (m) y = 4E-05x + 0.0285 y = 0.0007x - 0.2864 R² = 0.004 Transect 2 R² = 0.3648 0.3 p<0.05, df=13 Change in. A. germinans p=0.882 0.2 0.1 0 -0.1 0 100 200 300 400 500 600 700 -0.2 -0.3 Distance (m) y = 1E-05x - 0.0057 y = 0.0002x - 0.0747 R² = 0.0017 Transect 3 R² = 0.0403 0.2 p=0.924 p=0.492 Change in A. germinans 0.1 0 0 100 200 300 400 500 600 700 -0.1 -0.2 -0.3 Distance (m) y = 0.0005x - 0.027 y = -0.0003x + 0.1529 R² = 0.0158 Transect 4 R² = 0.1294 0.6 p=0.767 p=0.206 Change in A. germinans 0.4 0.2 -1E-15 -0.2 0 100 200 300 400 500 600 700 -0.4 -0.6 Distance (m)Figure 3.18a Regression curves for transects 1-4 for correlation between distance from road and change in relativeabundance of A. germinans in Atasta Lagoon from 1984 to 2009. Shaded are significant correlations. - 61 -
    • y = 0.001x - 0.1817 y = 5E-07x + 0.0175 R² = 0.2304 Transect 5 R² = 3E-06 0.3 p=0.229 p=0.995 Change in A. germinans 0.2 0.1 0 -0.1 0 100 200 300 400 500 600 700 -0.2 -0.3 Distance (m) y = 6E-05x - 0.007 Transect 6 y = -5E-05x + 0.0165 R² = 0.0131 R² = 0.0372 0.2 Change in A. germinans p=0.788 p=0.509 0.15 0.1 0.05 0 -0.05 0 100 200 300 400 500 600 700 -0.1 -0.15 -0.2 Distance (m) y = 0.0011x - 0.0721 Transect 7 y = -0.001x + 0.6292 R² = 0.4633 R² = 0.3183 p=0.063 p=0.056 Change in A. germinans 0.6 0.4 0.2 -1E-15 -0.2 0 100 200 300 400 500 600 700 -0.4 -0.6 Distance (m) y = -0.0009x + 0.1285 y = -0.0002x + 0.1164 Transect 8 R² = 0.4905 R² = 0.0071 0.6 Change in A. germinans p=0.775 0.4 0.2 0 0 100 200 300 400 500 600 700 -0.2 -0.4 Distance (m)Figure 3.18b Regression curves for transects 5-8 for correlation between distance from road and change in relative abundance of A. germinans in Atasta Lagoon from 1984 to 2009. - 62 -
    • 3.6.3 Change in Simpson’s biodiversity with distance from road There are no statistically or visibly significant correlations between distance from road andchange in Simpson’s biodiversity (1-D) in both divisions (Figure 3.19a-b). There is a general trend fornegative linear correlation for transects 1, 3, and 4 and positive linear correlation in transects 2 and 5 forthe first division (0-225m) (Figure 3.19a-b). After the 225m division, there is no visible trend betweendistance and change in biodiversity in all transects (Figure 3.19a-b). Transects 6, 7 and 8 demonstrate notrends for effect of the road on change in biodiversity, and transects 7 and 8 also include noise in thedata, like for the other test variables (Figure 3.19b). - 63 -
    • y = -0.0004x + 0.0468 Transect 1 0.3 y = 3E-05x - 0.0158 R² = 0.3062 R² = 0.0077 0.2 p=0.155 p=0.766 Change in 1-D 0.1 0 0 100 200 300 400 500 600 700 -0.1 -0.2 Distance (m) y = 0.0003x - 0.12 Transect 2 y = -0.0002x + 0.0041 R² = 0.0143 R² = 0.0081 p=0.778 p=0.760 0.2 Change in 1-D -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m) y = -0.0005x + 0.0441 Transect 3 y = -0.0004x + 0.169 R² = 0.0801 R² = 0.0647 0.4 p=0.497 p=0.380 Change in 1-D 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m) y = -0.0007x - 0.1398 Transect 4 y = 0.0003x - 0.2017 R² = 0.0314 0.4 R² = 0.013 p=0.675 p=0.698 Change in 1-D 0.2 -1E-15 0 100 200 300 400 500 600 700 -0.2 -0.4 -0.6 Distance (m)Figure 3.19a Regression curves for transects 1-4 for correlation between distance from road and change in biodiversity of mangroves in Atasta Lagoon from 1984 to 2009. - 64 -
    • y = 0.0007x - 0.1049 Transect 5 y = 0.0001x - 0.0793 0.3 R² = 0.1892 R² = 0.1100 p=0.281 p=0.247 0.2 Change in 1-D 0.1 0 -0.1 0 100 200 300 400 500 600 700 -0.2 -0.3 -0.4 Distance (m) y = -0.0014x + 0.1629 Transect 6 y = 0.0004x - 0.1695 0.3 R² = 0.1126 R² = 0.1799 p=0.417 p=0.131 0.2 Change in 1-D 0.1 0 -0.1 0 100 200 300 400 500 600 700 -0.2 -0.3 -0.4 Distance (m) y = -0.0014x + 0.3151 Transect 7 y = 0.0012x - 0.763 R² = 0.079 R² = 0.1155 0.6 p=0.280 p=0.500 0.4 Change in 1-D 0.2 0.0 -0.2 0 100 200 300 400 500 600 700 -0.4 -0.6 Distance (m) y = 0.0023x - 0.366 Transect 8 y = 0.0004x - 0.1437 0.3 R² = 0.3107 R² = 0.0803 0.2 p=0.250 p=0.326 0.1 Change in 1-D -1E-15 -0.1 0 100 200 300 400 500 600 700 -0.2 -0.3 -0.4 -0.5 -0.6 Distance (m)Figure 3.19b Regression curves for transects 5-8 for correlation between distance from road and change in biodiversity of mangroves in Atasta Lagoon from 1984 to 2009. - 65 -
    • 4.0 DISCUSSION4.1 Effect of road on R. mangle and A. germinans Based on the ANOVA analyses for distance-effect correlations across the 8 digital transects, theresults suggest that the road negatively impacts the relative abundance of R. mangle for the 0-225msections in 2 transects, and the 255-645m section for 1 transect, in partial support of the first hypothesisfor this study. The results do not support the second hypothesis that the road negatively impacts therelative abundance of A. germinans. There are two positive correlations for the 255-645m sections oftwo transects, but the lack of evidence for a correlation with proximity to the road suggests either thatthere is no effect or that there is a substantial amount of noise in the data transmission from fieldcollection and regression with the SVI to comparison with distance using ArcGIS. There is no evidence of competition between the two dominant species, R. mangle and A.germinans for the areas sampled in the field. Only riverine sections of the forest were sampled, alongthe border of the lagoon, so this reflects only the dynamics occurring for these areas, where bothspecies can be found in mutually high abundances (Figure 2.5). The lack of relationship betweenabundance of either of the dominant species and Simpson’s biodiversity confirms that the index is anindication of evenness of species, not relative dominance of one (Hill, 1973). Theoretically, the construction of the road would have three direct consequences on thesurrounding mangroves: modification of soil conditions, increase in light availability from gapintroductions and obstruction of water flow. Constructed pathways through mangrove forests have hadunforeseen and long-term indirect impacts in the past. A boardwalk constructed through mangroves(Avicennia marina) in Australia modified macrofauna assemblages up to 24m from the boardwalk(Kelaher et al. 1998a). With increasing distance from the boardwalk, there was also an increase in - 66 -
    • pneumatophores density for this species, and crab species abundance was high near the boardwalk(Kelaher et al. 1998b). These indirect impacts of a linear anthropogenic disturbance are consequences ofsediment changes during construction as well as increased light exposure to the soil (Kehalher et al.1998a, 1998b). The road would have created substantial gaps along the periphery of the elevated soil and watertable, making the mangroves found along the edge exposed to “edge effects”. In Rhizophora sp. andAvicennia sp., exposure to an increased amount of sunlight leads to structural consequences in growth:gnarled and lateral branching as well as increased root density (Duke, 2001). Under natural lightconditions, after the maturing phase in the mangrove, a thinning typically occurs, followed bysenescence after a given amount of years (Duke, 2001). Given the increased exposure to light that theroads induced, the mangroves likely allocated their energy to lateral branching and increased rootdensity, accelerating the senescence phase (Duke, 2001). R. mangle was once thought as completelyshade-tolerant, but numerous studies have demonstrated that under varying light conditions, thisspecies is actually able to persist and re-establish itself following canopy gap generation and has beensuggested as being dependent on these gaps for persistence (Chen and Twilley, 1998). Koch (1997)found that growth rates in R. mangle were 2-5x greater in canopy gaps created by hurricanes than inclosed canopies in the Everglades, Florida. Community competition in mangrove forests has been modeled based on neighbourhooddynamics. The ‘field of neighbourhood’ approach proposed by Berger and Hildenbrandt (2000) treatsmangroves as separate entities with overlapping ‘zones of influence’, which describe the tree spacingeffects that can be modelled to predict spatial patterns for R. mangle and A. germinans. The modelsdescribed conclude that if both species establish in an area at the same time, R. mangle dominates thestand due to a faster growth rate, but is outcompeted by A. germinans decades later due to the longerlife span in this species (Berger and Hildenbrandt, 2000). If one species establishes in an area, followed - 67 -
    • by the other species, the pioneer will always maintain dominance in the area (Berger and Hildenbrandt,2000). It is unknown which species was first established in the Atasta Peninsula. If both species enteredand established in the lagoon, based on the logic proposed by this model, A. germinans has had a verylong period of time to become the dominant species in the area, and from which, R. mangle may besuppressed along the highway due to advantages taken by A. germinans with the increased availabilityof light. Increased light is just one consequence of gap formation caused by mangrove deforestation. Inthis case, elevation of the soil and water table would modify hydrology dynamics and soil conditionssurrounding the constructed road. Constriction of water flow may cause soil salinity to increase alongthe periphery of the road, as the lagoon is brackish (NaCl concentration is lower than ocean water, buthigher than freshwater). A. germinans, with a high growth rate and high tolerance to high salinityconditions, may have a competitive advantage over other species in this way (Lovelock and Feller, 2003). - 68 -
    • 4.2 Effect of road on Simpson’s biodiversity Based on the ANOVA analyses for distance-effect correlations across the 8 digital transects, theresults suggest that the road has no effect on Simpson’s biodiversity (1-D) of the four mangrove species,supporting the third null hypothesis. Given that the relationship between field-collected and SVI-predicted values for this test variable was very strong, it is ample to state that the results suggest thatthe road has no impact on overall evenness of mangrove species in the area. Noise from datatransmission in this case is likely very minimal, when compared to that of the prediction of relativeabundance of A. germinans. The overall decrease in this variable by 0.03 units (4.6% from 1984 value) for the 2,660ha studyarea should be considered with the units expressed using the Simpson’s biodiversity index (1-D) in mind.The range for 1-D is 0.0 for no biodiversity to 1.0 for complete evenness of species, so a 4.6% decreasefrom estimated values for 1984 is substantial. Mangroves are steady-state ecosystems; they have the ability to persist through naturalstressors via positive and negative feedbacks. The mangrove ecosystem as a whole experiences stress ona cycle; when stress is low in the absence of disturbance for a long period of time, species diversity ishighest, but is regulated by future periods of stress caused by disturbance (Lugo, 1980). Human-createdstressors, caused by disturbances such as the constructed road here, by comparison, create generalpatterns of perturbation that cause an overall decrease in height and biomass of the mangroves,resulting in a potential decrease in biodiversity over time (Lugo, 1980). The study period used here is 25years. It is possible (and most likely) that the disturbance caused by the road poses a threat to overallbiodiversity if intensity and frequency is high, and may only be detectable after a long period of time,after the senescence stage. - 69 -
    • 4.3 Efficacy of SVI in prediction of test variables The use of remote sensing for mangrove studies has come a long way in recent years, fromaerial photography to Landsat MSS and TM to high resolution sensors like SPOT, ASTER or IRS C and D(Heumann, 2011). All methods still have their applicability for mangrove studies, depending on thescope and scale of the research. Rhizophora sp. reflect red and infrared energy approximately 10% more than Avicennia sp.,making it possible to discriminate generalized patterns of relative density in mangrove forests (Blasco etal. 1998). Due to atmospheric effects observed in satellite data, especially for areas where humidity ishigh. All of the SVI selected include one or more factors that correct for atmospheric effects. Theselections for suitable SVI to predict the test variables were a result of a series of regressions forgoodness-of-fit and not extensive testing for their efficacy in predicting the field variables. In light ofthis, the statistically significant correlations found for predicting relative abundance of R. mangle andSimpson’s biodiversity (1-D) were both strong enough to apply to the whole area for hypothesis testingin this study. The Normalized Difference Water Index (NDWI), also named the Infrared Index (IRI), was used inthis study to predict relative abundance of R. mangle. The relationship between NDWI and abundanceof R. mangle in this study is not the tightest fit as anticipated, but in terms of predicting overall patternsof R. mangle abundance for the area, it serves its purpose. The index is given as the subtraction ofwavelength 1.24µm by 0.86µm and divided by the addition of these two wavelengths (Gao, 1996). Thesewavelengths correspond to the near infrared (Band 4) and mid-infrared (Band 5) wavelengths given inLandsat TM data ([TM4-TM5] / [TM4+TM5]) (Jackson et al. 2004).Mid-infrared radiation is not absorbedby plants and has a lower leaf transmission, so scattering is less apparent than with near-infarredradiation (Steininger, 2000). The index has been effectively used to predict and monitor changes in cornand soybean vegetative water content (VWC), even after NDVI, one of the most widely used indices - 70 -
    • ([TM4-TM3] / [TM4+TM3]), saturated and failed to detect changes (Jackson et al. 2004). NDWI is alsoless sensitive to atmospheric effects than NDVI (Gao, 1996). The index has also been associated with dryplant above-ground biomass in tropical forests of Brazil (R2>0.64, p<0.01) (Steininger, 2000) and relativetree abundance in a semi-deciduous tropical dry forest in Venezuela (rs=−0.55, p<0.01) (Feeley et al.2005). The prediction of A. germinans by GEMI is not as strong as for the other two test variables, withthe lack of statistical significance. GEMI is an index with a complex nonlinear formula, unlike the otherSVI evaluated, with several factors using bands 3 and 4 as independent variables. Bands 3 and 4,displaying red and infrared reflectance from the top of the atmosphere, are useful to use in combinationwith one another, due to the low reflectance of energy in visible wavelengths of light and highreflectance in the infrared (Pinty and Verstraete, 1992). The basic formula for comparing these bands isTM4/TM3 (the Simple Ratio), which has been modified over time with testing to produce ratios withhigher complexity like NDVI and GEMI (Table 2.2) (Pinty and Verstraete, 1992). GEMI was proposed toovercome the effects of atmospheric and illumination effects as well as bias to one vegetation type(Pinty and Verstraete, 1992). In the case of the study presented here, the cubic trend observed betweenreflected GEMI values and measured abundance of A. germinans is visibly apparent, but not statisticallystrong enough to draw conclusions based on their correlation. The quadratic relationship for prediction of Simpson’s biodiversity (1-D) of mangroves by EVI inthis study is very strong, relative to the two other variables evaluated. The Enhanced Vegetation Index(EVI), also named the Soil And Atmosphere Resistant Vegetation Index (SARVI2), utilizes bands 1, 3 and 4in its formula, for visible blue, visible red and visible infrared radiation. The combination of the threebands is a modification of NDVI, with a soil adjustment factor and inclusion of the blue band forreduction of atmospheric aerosol scattering produced by the red wavelength band (Jenson, 2000). Themodification was made to remove canopy background noise for the Moderate Resolution Imaging - 71 -
    • Spectroradiometer (MODIS) platform on the Earth Observing Satellite (EOS) (Huete et al. 1997). MODIShas similar band structure to those of the Landsat TM sensor and can be calibrated with Landsat TMimages for analysis, so the index was consequently used in this study. When used to assess forest typesaround the world (ex. Siberian boreal forest, Amazon Basin tropical forests) using MODIS and LandsatTM data, there was a strong positive correlation between field-measured near-infrared reflectance inthe forests and EVI value displayed in the images, but no relationship was found with measured redreflectances (Huete et al. 1997). EVI also demonstrated a strong relationship with leaf area index (LAI) inthe forests (Huete et al. 1997). - 72 -
    • 4.4 Limitations of study Accurate discrimination of individual mangroves to the species level is still extremely difficult,even with the highest spatial and spectral resolution. Mixed pixels have been demonstrated as a primarysource of error in attempts to estimate the leaf-area-index of mangroves with NDVI using Landsat TMdata (Green et al. 1997). In attempt to estimate mangrove biomass images in The Guangdong Provincein South China using NDVI and Radarsat, it was concluded that Radarsat has advantages over theLandsat TM index NDVI, due to the index’s overestimate of certain species and underestimate of others(Li et al. 2007). In cases like this, error may arise due to high biomass or density of mangroves in an area,which corroborates also with the dense mangroves evaluated in this study. Challenges like these limitthe accuracy of results when using 30m resolution data. The tropics present a myriad of potential errors that must be considered and accounted for inremote sensing studies. While high resolution remote sensing technology is invaluable in tropical andcoastal management, its application in quantitative analysis for ecology studies is still less than perfect.The integration of high resolution, hyperspectral remote sensing technology with aerial photographyand GIS make optimizes analysis of forest stand characteristics (Dahdough-Guebas, 2002). For the scopeof this study, the cost for spatial data and timeline needed for this type of analysis were not available. - 73 -
    • 5.0 CONCLUSIONS AND RECOMMENDATIONS Roads constructed in forested areas have many consequences on the ecosystem. Fragmentationof a forest can have consequences for wildlife continuity, hydrology aspects and reproductivepersistence of trees in the forest. In addition, constructed roads have consequences for land usechanges, especially in the tropics where mangroves are used for timber. The results presented here are preliminary assessments for evidence of change in the dominantspecies and overall biodiversity of mangroves. A new road is planned for the area, to make a connectionbetween Atasta and Palizada (Bach et al. 2005). The road would be placed through the wetlandssurrounding the Términos Lagoon, which is also fringed with mangroves (Bach et al. 2005).Environmentalists in the area are concerned about habitat fragmentation, constricted hydrologydynamics and increased urban traffic that the road would cause. This research may have greater insightinto how to test for and monitor large-scale changes in the species patterns and overall biodiversity ofmangroves using Landsat TM imagery. Sensors with higher resolution (Quickbird, SPOT, Ikonos, CASI)can also be used to address these issues, as well as intensive empirical research concerning hydrologydynamics surrounding roads. For this study, the conclusions made based on the data provided by field analysis integratedwith multispectral satellite image analysis and GIS should be considered as preliminary findings thatsuggest a general trend for the described impacts of the road. Empirical studies are encouraged tosubstantiate the suggestions made by this study in order to quantify the degree of the described impactscaused by the road. - 74 -
    • 6.0 REFERENCESAlongi D.M. 2002. Present state and future of the world’s mangrove forests. Environmental Conservation 29: 331-349.Alongi, D.M. 2008. Mangrove forests: resilience, protection from tsunamis, and responses to global climate change. Estuarine, Coastal and Shelf Science 76: 1-13.Aplin P. 2005. Remote sensing: ecology. Progress in Physical Geography 29: 104-113.Bach L., Calderon R., Cepeda M. F., Oczkowsk A., Olsen S.B. and Robadue, D. 2005. Level One Site Profile: Laguna de Términos and its Watershed, Mexico. Narragansett, RI: Coastal Resources Center, University of Rhode Island. 29pp.Barbier E.B. 2000. Valuing the environment as input: review of applications to mangrove-fishery linkages. Ecological Economics 35: 47-61.Barreiro-Güemes M.T. 1999. Aporte de hojarasca y renovación foliar del manglar en un sistema estuarino del Sureste de México. Revista de Biología Tropical 47: 729-737.Baugh W.M. and Groeneveld D.P. 2006. Broadband vegetation index performance evaluated for a low- cover environment. International Journal of Remote Sensing 27: 4715-4730.Béland M. , Goïta K. , Bonn F. and Pham T. T. H. 2006. Assessment of land-cover changes related to shrimp aquaculture using remote sensing data: a case study in the Giao Thuy District, Vietnam. International Journal of Remote Sensing, 27: 8, 1491-1510.Berger U. and Hildenbrandt H. 2000. A new approach to spatially explicit modelling of forest dynamics: spacing, ageing and neighbourhood competition of mangrove trees. Ecological Modelling 132: 287-302.Berger U., Adams, M, Grimm V. and Hildenbrandt H. 2006. Modelling secondary succession of neotropical mangroves: causes and consequences of growth reduction in pioneer species. Perspectives in Plant Ecology, Evolution and Systematics 7: 243-252.Blasco F., Gauquelin T., Rasolofoharinoro M., Denis J., Aizpuru M. and Caldairou V. 1998. Recent advances in mangrove studies using remote sensing data. Marine Freshwater Research 49: 287- 296.Cabrera M. A., Seijo J. C., Euan, J. and Perez, E., 1998. Economic values of ecological services from a mangrove ecosystem. International Newsletter of Coastal Management 33: 1-2.Campbell J.B. 2007. Introduction to remote sensing: 4th edition. The Guilford Press, New York, NY. ISBN: 978-1-60623-074-9. 626 pp. - 75 -
    • Chen R. and Twilley R.R. 1998. A gap dynamic model of mangrove forest development along gradients of soil salinity and nutrient resources. Journal of Ecology 86: 37-51.CONABIO. 2009. Manglares de México: Extensión y distribución. 2nd ed. Comisión Nacional para el Conocimiento y Uso de la Biodiversidad. México. 99 pp. Available online at: http://www.conabio.gob.mx/conocimiento/manglares/doctos/Manglares_de_Mexico_Extension _y_distribucion.pdfDiorio Mimi , Jupiter S.D. , Cochran S.A. and Potts D.C. 2007. Optimizing remote sensing and GIS tools for mapping and managing the distribution of an invasive mangrove (Rhizophora mangle) on South Molokai, Hawaii. Marine Geodesy 30: 125-144.Dahdough-Guebas F. 2002. The use of remote sensing and GIS in the sustainable management of tropical coastal ecosystems. Environment, Development and Sustainability 4: 93-112.Davis B.A. and Jensen J.R. 1998. Remote sensing of mangrove biophysical characteristics. Geocarto International 13: 55-64.Díaz B.M. and Blackburn G.A. 2003. Remote sensing of mangrove biophysical properties: evidence from a laboratory simulation of the possible effects of background variation on spectral vegetation indices. International Journal of Remote Sensing 24: 53-73.Duke N.C. 2001. Gap creation and regenerative processes driving diversity and structure of mangrove ecosystems. Wetlands Ecology and Management 9: 257-269.Ewel K.C., Twilley R.R. and Ong J.E. 1998. Different kinds of mangrove forests provide different goods and services Global Ecology and Biogeography Letters 7: 83-94.Ellison A.M. and Farnsworth E.J. 1993. Seedling survivorship, growth and response to disturbance in Belizean mangal. American Journal of Botany 80: 1137-1145.Ellison A.M. and Farnsworth E.J. 1996. Anthropogenic disturbance of Caribbean mangrove ecosystems: past impacts, present trends and future predictions. Biotropica 28: 549-565.Everitt J.H. and Judd F.W. 1989. Using remote sensing techniques to distinguish and monitor black mangrove (Avicennia germinans). Journal of Coastal Research 5: 737-745.Farnsworth E.J. and Ellison A.M. 1996. Sun-shade adaptability of the red mangrove, Rhizophora mangle (Rhizophoraceae): changes through ontogeny at several levels of biological organization. Journal of Botany 83: 1131-1143.Farnsworth E.J. and Ellison A.M. 1997. The global conservation status of mangroves. Ambio 26: 328-334.FAO. 2007. The worlds mangroves 1980-2005. FAO Forestry Paper No. 153. Rome. 83pp.Feeley K.J., Gillespie T.W. and Terborgh J.W. 2005. The utility of spectral indices from Landsat ETM+ for measuring the structure and composition of tropical dry forests. Biotropica 37: 508-519. - 76 -
    • Gao B. 1996. NDWI - A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sensing of the Environment 58: 257-266.Giri C., Ochieng E., Tieszen L.L., Zhu Z., Singh A., Loveland T., Masek J. and Duke N. 2010. Status and distribution of mangrove forests of the world using earth observation satellite data. Global Ecology and Biogeography. DOI: 10.1111/j.1466-8238.2010.00584.xGould W. 2000. Remote sensing of vegetation, plant species richness, and regional biodiversity hotspots. Ecological Applications 10: 1861-1870.Green E.P., Mumby P.J., Edwards A.J., Clark C.D. and Ellis A.C. 1997. Estimating leaf area index of mangroves from satellite data. Aquatic Botany 58: 11-19.Hill M.O. 1973. Diversity and evenness: a unifying notation and its consequences. Ecology 54: 427-432.Heumann B.W. 2011. Satellite remote sensing of mangrove forests: Recent advances and future opportunities. Progress in Physical geography 35: 87-108.Huete A.R. Liu H.Q., Batchily K. and van Leeuwen W. 1997. A comparison of vegetation indices over a global set of TM images for EOS-MODIS. Remote Sensing of the Environment 59: 440-451.Implan. 2010. Monografias – Atasta. Instiututo municipal de planeación de Carmen. Available at: [http://www.implancarmen.org/pdf/10/Atasta.pdf].Jackson T.J., Chen D., Cosh M., Li F., Anderson M., Walthall C., Doriaswamy P. and Hunt E.R. 2004. Vegetation water content mapping using Landsat dara derived normalized different water index for corn and soybeans. Remote Sensing of the Environment 92: 475-482.Jensen J.R. 2000. Remote sensing of vegetation.p. 333-377. In: Remote sensing of the environment, an earth resource perspective. Upper Saddle River,NJ: Prentice Hall. 544 pp.Kaufman Y.J. and Tanré D. 1996. Strategy for direct and indirect methods correcting the aerosol effect on remote sensing: from AVHRR to EOS-MODIS. Remote Sensing of the Environment 55: 65- 79.Kathiresan K. and Bingham G.L. 2001. Biology of Mangroves and Mangrove Ecosystems. Advances in Marine Biology 40: 81-251.Kelaher B.P., Chapman M.G. and Underwood A.J. 1998a. Changes in benthic assemblages near boardwalks in temperate urban mangrove forests. Journal of Experimental Marine Biology and Ecology 228: 91-307.Kelaher B.P., Underwood A.J. and Chapman M.G. 1998b. Effect of boardwalks on the semaphore crab Heloecius cordiformis in temperate urban mangrove forests. Journal of Experimental Marine Biology and Ecology 227: 281-300.Kerr J.T. and Ostrovsky M. 2003. From space to species: ecological applications for remote sensing. Trends in Ecology and Evolution 18: 299-305. - 77 -
    • Kovacs J.M., Wang J. and Blanco-Correa M. 2001. Mapping disturbances in a mangrove forest using multi-date Landsat TM imagery. Environmental Management 27: 763-776.Krauss K.W., Lovelock C.E., McKee K.L., Lopez-Hoffman L., Ewe S.M.L. and Sousa W.P. 2008. Environmental drivers in mangrove establishment and early development: A review. Aquatic Botany 89: 105-127.Lee T. and Yeh H. 2009. Applying remote sensing techniques to monitor shifting wetland vegetation: a case study of Danshui River estuary mangrove communities, Taiwan. Ecological Engineering 35: 487-496.Long B.G. and Skewes T.D. 1996. A technique for mapping mangroves with Landsat TM satellite data and Geographic Information System. Estuarine, Coastal and Shelf Science 43: 373-381.López-Portillo J. and Ezcurra E. 1989. Response of three mangroves to salinity in two geoforms. Functional Ecology 3: 355-361.López-Portillo J. and Ezcurra E. 2002. Los manglares de México: una revision. Madera y Bosques Número especial: 27-51.Lovelock C.E. and Feller I.C., 2003. Photosynthetic performance and resource utilization of two mangrove species coexisting in hypersaline scrub forest. Oecologia 134: 455–462.Lugo A.E. and Snedaker S.C. 1974. The ecology of mangroves. Annual Review of Ecology and Systematics 5: 39-64.Lugo A.E. 1980. Mangrove ecosystems: successional or steady state? Biotropica 12: 65-72.Mas J.F. 2004. Mapping land use/cover in a tropical coastal area using satellite sensor data, GIS and artificial neural networks. Estuarine, Coastal and Shelf Science 59: 219-230.Mironga J.M. 2004. Geographic Information Systems (GIS) and remote sensing in the management of shallow tropical lakes. Applied Ecology and Environmental Research 2: 83-103.Myeni R.B., Hall F.G., Sellers P.J. and Marshak A.L. 1995. The interpretation of spectral vegetation indexes. IEEE Transactions on Geoscience and Remote Sensing 33: 481-486.Pinty B. and Verstraete M.M. 1992. GEMI: a non-linear index to monitor global vegetation from satellites. Vegetatio 101: 15-20.Ramírez-García P., López-Blanco J. and Ocana D. 1998. Mangrove vegetation assessment in the Santiago River Mouth, Mexico, by means of supervised classification using Landsat TM imagery. Forest Ecology and Management 105: 217-229.Ruiz-Marín A., Campos-Garcia S., Zavala-Loría J. and Canedo-Lopez Y. 2009. Hydrological aspects of the lagoons Atasta and Pom, Mexico. Tropical and Subtropical Agroecosystems 10: 63-74. - 78 -
    • Sherman R.E., Fahey T.J. and Battles J.J. 2000. Small-scale disturbance and regeneration dynamics in a neotropical mangrove forest. Journal of Ecology 88: 165-178.Sims D.A. and Gamon J.A. 2002. Relationships between leaf pigment content and spectral reflectance across a wide range of species, leaf structures and developmental stages. Remote Sensing of Environment 81: 337-354.Soto-Galera E., Piera J. and López P. 2010. Spatial and temporal land cover changes in Términos Lagoon Reserve, Mexico. International Journal of Tropical Biology 58: 565-575.Sousa W.P. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15: 353-391.Sulong I., Mohd-Lokman H., Mohd-Tarmizi K. and Ismail A. 2002. Mangrove mapping using Landsat imagery and aerial photographs: Kehaman District, Terengganu, Malaysia. Environment, Development and Sustainability 4: 135-152.Todd S.W. and Hoffer R.M. 1998. Responses of spectral indices to variations in vegetation cover and soil background. Photogrammetric Engineering & Remote Sensing 64: 915-921.Vega, M.A. 2005. Plan de Conservación para la Reserva de la Biosfera Pantanos de Centla y el Área de protección de flora y fauna Laguna de Términos. PRONATURA Península de Yucatán/Programa Costero, México. 132 pp.Vincini M., Frazzi E. and D’Alesio P. 2008. A broad-band leaf chlorophyll vegetation index at the canopy scale. Precision Agriculture 9: 303-319.Xie Y., Sha Z. And Yu M. 2008. Remote sensing imagery in vegetation mapping: a review. Journal of Plant Ecology 1: 9-23.Yáñez-Arancibia A., Lara-Domínguez A.L., Rojas Galavis J.L., Zárate Lomeli D.J.,Villalobos Zapata G.J. and Sánchez-Gil P. 1999. Integrating science and management on coastal marine protected areas in the Southern Gulf of Mexico. Ocean & Coastal Management 42: 319-344.Yanagisawa H., Koshimura S., Goto K., Miyagi T., Imamura F., Ruangrassamee A. and Tanavud C. 2009. The reduction effects of mangrove forest on a tsunami based on field surveys at Pakarang Cape, Thailand and numerical analysis. Estuarine, Coastal and Shelf Science 81: 27-37. - 79 -
    • 7.0 APPENDICES Appendix A: Raw field dataTable 7.1 Abundance counts of mangrove in 30x30m sampling plots. Quadrat 5 was omitted from analysis due to inaccessibility for high density of mangroves.Quadrat Easting Northing Longitude Latitude R. mangle A. germinans L. racemosa C. erectus L. Simpson’s (m) (m) (degrees, minutes, (degrees, minutes, index (1-D) seconds) seconds) 1 599280 2059050 -92°03’31.9337 18°37’12.2438 61 2 12 0 0.326 2 599310 2058870 -92°03’30.9422 18°37’06.3828 55 40 38 0 0.662 3 599430 2058630 -92°03’26.8905 18°36’58.5545 10 54 41 0 0.579 4 601530 2057550 -92°02’15.4317 18°36’23.0569 28 36 0 6 0.576 6 601410 2056950 -92°02’19.6357 18°36’03.5583 47 24 4 2 0.534 7 599880 2056170 -92°03’11.9781 18°35’38.4473 56 32 15 1 0.600 8 599790 2056050 -92°03’15.0702 18°35’34.5588 59 56 42 0 0.664 9 599910 2055840 -92°03’11.0139 18°35’27.7064 60 40 47 0 0.662 10 601050 2054340 -92°02’32.3930 18°34’38.7110 45 74 78 5 0.670 11 601170 2054850 -92°02’28.2065 18°34’55.2818 125 85 33 0 0.597 12 601620 2054490 -92°02’12.9205 18°34’43.4919 96 87 84 10 0.690 13 602040 2055630 -92°01’58.3832 18°35’20.5055 48 37 62 10 0.695 14 602400 2055570 -92°01’46.1123 18°35’18.4905 125 76 0 0 0.473 15 602578 2055510 -91°01’40.0506 18°35’16.5072 105 69 66 2 0.659 16 603270 2058480 -92°01’15.8900 18°36’53.0059 46 43 0 3 0.536 17 603480 2058570 -92°01’08.7077 18°36’55.8965 0 51 0 0 0.000 80
    • Appendix B: ANOVA results in curve fitting for candidates for suitable SVI for predicting relative abundance of R. mangle. Analyses processed by SPSS Statistics 17.0. Selected SVI is highlighted.Table 7.2 Results of curve fitting for TM4 prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .07587 1.14935 1 14 .30182 .121 .004 Logarithmic .06317 .94395 1 14 .34775 -.671 .256 Inverse .04957 .73018 1 14 .40721 .631 - 14.660 Quadratic .11681 .85966 2 13 .44603 1.161 -.027 .000 Cubic .11311 .82900 2 13 .45830 .800 -.011 .000 .000 Compound . . . . . .000 .000 Power . . . . . .000 .000 S . . . . . .000 .000 Growth . . . . . .000 .000 Exponential . . . . . .000 .000 Logistic . . . . . .000 .000Table 7.3 Results of curve fitting for TM5 prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .03254 .47083 1 14 .50381 .587 -.006 Logarithmic .03748 .54521 1 14 .47248 1.016 -.184 Inverse .04254 .62199 1 14 .44346 .219 5.060 Quadratic .04787 .32683 2 13 .72696 1.178 -.051 .001 Cubic .04787 .32683 2 13 .72696 1.178 -.051 .001 .000 Compound . . . . . .000 .000 Power . . . . . .000 .000 S . . . . . .000 .000 Growth . . . . . .000 .000 Exponential . . . . . .000 .000 Logistic . . . . . .000 .000Table 7.4 Results of curve fitting for TM7 prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00002 .00032 1 14 .98607 .416 .000 Logarithmic .00049 .00693 1 14 .93482 .448 - .017 Inverse .00146 .02048 1 14 .88824 .384 .217 Quadratic .01866 .12363 2 13 .88474 .809 - .006 .101 Cubic .02295 .15268 2 13 .85992 .567 .000 - .001 .007 Compound . . . . . .000 .000 Power . . . . . .000 .000 S . . . . . .000 .000 Growth . . . . . .000 .000 Exponential . . . . . .000 .000 Logistic . . . . . .000 .000 81
    • Table 7.5 Results of curve fitting for NDVI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .01209 .17129 1 14 .68524 .225 .289 Logarithmic .01211 .17161 1 14 .68496 .493 .184 Inverse .01228 .17402 1 14 .68289 .594 - .117 Quadratic .01228 .08082 2 13 .92282 .433 - .510 .367 Cubic .01298 .08548 2 13 .91858 .456 .000 - 1.056 .805 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.6 Results of curve fitting for SIPI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .06508 .97450 1 14 .34032 .344 .252 a Logarithmic . . . . . .000 .000 Inverse .01212 .17179 1 14 .68481 .411 - .002 Quadratic .08580 .61005 2 13 .55817 .342 - .732 .040 Cubic .09733 .43131 3 12 .73439 .303 - 3.072 - .275 3.486 b Compound . . . . . .000 .000 a,,b Power . . . . . .000 .000 b S . . . . . .000 .000 b Growth . . . . . .000 .000 b Exponential . . . . . .000 .000 b Logistic . . . . . .000 .000Table 7.7 Results of curve fitting for SR prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .01525 .21682 1 14 .64863 .320 .019 Logarithmic .01305 .18509 1 14 .67359 .278 .086 Inverse .01204 .17059 1 14 .68585 .496 - .387 Quadratic .03124 .20962 2 13 .81358 .717 - .015 .143 Cubic .04943 .33799 2 13 .71929 .557 .000 - .003 .021 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000 82
    • Table 7.8 Results of curve fitting for NDWI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .32740 6.81488 1 14 .02055 -.389 1.841 Logarithmic .34010 7.21537 1 14 .01773 1.127 .851 Inverse .34940 7.51858 1 14 .01589 1.308 -.383 Quadratic .35212 3.53271 2 13 .05953 -1.848 8.291 - 6.987 Cubic .35212 3.53271 2 13 .05953 -1.848 8.291 - .000 6.987 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.9 Results of curve fitting for GVI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00532 .07493 1 14 .78828 .357 .012 Logarithmic .00357 .05018 1 14 .82599 .342 .048 Inverse .00205 .02881 1 14 .86764 .452 - .166 Quadratic .01668 .11029 2 13 .89641 .735 - .016 .149 Cubic .01691 .11182 2 13 .89506 .507 .000 - .002 .015 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.10 Results of curve fitting for CVI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00520 .07317 1 14 .79072 .322 .041 Logarithmic .00216 .03029 1 14 .86433 .368 .056 Inverse .00038 .00537 1 14 .94260 .435 -.048 Quadratic .05643 .38876 2 13 .68552 1.632 - .277 1.179 Cubic .05759 .39719 2 13 .68009 .829 .000 - .087 .284 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000 83
    • Table 7.11 Results of curve fitting for MIRI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .01569 .22318 1 14 .64391 .496 -.023 Logarithmic .01654 .23543 1 14 .63503 .529 -.094 Inverse .01987 .28380 1 14 .60258 .300 .377 Quadratic .01973 .13085 2 13 .87849 .342 .056 -.009 Cubic .49387 3.90312 3 12 .03704 7.336 - 1.358 - 5.450 .107 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.12 Results of curve fitting for GEMI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .07353 1.11117 1 14 .30968 .298 .000 a Logarithmic . . . . . .000 .000 Inverse .02377 .34092 1 14 .56859 .464 125.439 Quadratic .10728 .78111 2 13 .47825 .451 .000 .000 Cubic .12712 .58251 3 12 .63774 .196 .000 .000 .000 b Compound . . . . . .000 .000 a,,b Power . . . . . .000 .000 b S . . . . . .000 .000 b Growth . . . . . .000 .000 b Exponential . . . . . .000 .000 b Logistic . . . . . .000 .000Table 7.13 Results of curve fitting for EVI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .01360 .19303 1 14 .66711 .095 .147 Logarithmic .01495 .21254 1 14 .65186 .159 .331 Inverse .01649 .23470 1 14 .63555 .758 -.742 Quadratic .04002 .27096 2 13 .76685 -6.824 6.613 - 1.503 Cubic .04002 .27096 2 13 .76685 -6.824 6.613 - .000 1.503 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000 84
    • Table 7.14 Results of curve fitting for TM4-TM2 prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .05862 .87185 1 14 .36627 .223 .004 Logarithmic .03927 .57225 1 14 .46191 -.133 .142 Inverse .02177 .31159 1 14 .58552 .508 - 4.314 Quadratic .12598 .93691 2 13 .41676 .881 -.026 .000 Cubic .12420 .92182 2 13 .42230 .679 -.012 .000 .000 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.15 Results of curve fitting for ARVI prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .13515 2.18774 1 14 .16126 .796 - .181 Logarithmic .12472 1.99490 1 14 .17967 .680 - .364 Inverse .11105 1.74888 1 14 .20722 .072 .697 Quadratic .14710 1.12107 2 13 .35550 .170 .421 - .140 Cubic .14360 1.08993 2 13 .36508 .439 .078 .000 - .018 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.16 Results of curve fitting for TM4-TM3 prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .06094 .90852 1 14 .35668 .213 .004 Logarithmic .04767 .70075 1 14 .41659 -.249 .166 Inverse .03498 .50742 1 14 .48796 .547 - 6.945 Quadratic .10175 .73633 2 13 .49782 .803 -.019 .000 Cubic .10493 .76197 2 13 .48650 .489 .000 .000 .000 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000 85
    • Table 7.17 Results of curve fitting for TM3-TM2 prediction of relative abundance of R. mangle. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .03079 .44482 1 14 .51565 .300 - .017 a Logarithmic . . . . . .000 .000 Inverse .02900 .41818 1 14 .52831 .511 .596 Quadratic .04182 .28366 2 13 .75757 .551 .066 .006 Cubic .06128 .42433 2 13 .66295 .482 .000 - - .009 .001 b Compound . . . . . .000 .000 a,,b Power . . . . . .000 .000 b S . . . . . .000 .000 b Growth . . . . . .000 .000 b Exponential . . . . . .000 .000 b Logistic . . . . . .000 .000 86
    • Appendix C: ANOVA analysis in curve fitting for candidates for suitable SVI for predicting relative abundance of A. germinans. Analyses processed by SPSS Statistics 17.0. Selected SVI is highlighted.Table 7.18 Results of curve fitting for TM4 prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .20774 3.67091 1 14 .07602 .864 -.007 Logarithmic .21203 3.76714 1 14 .07268 2.384 -.475 Inverse .21240 3.77552 1 14 .07239 -.082 30.714 Quadratic .21096 1.73785 2 13 .21436 1.159 -.016 .000 Cubic .21096 1.73785 2 13 .21436 1.159 -.016 .000 .000 Compound .22822 4.13996 1 14 .06128 2.056 .973 Power .20796 3.67595 1 14 .07584 460.185 -1.724 S .18705 3.22124 1 14 .09430 -2.727 105.582 Growth .22822 4.13996 1 14 .06128 .721 -.027 Exponential .22822 4.13996 1 14 .06128 2.056 -.027 Logistic .22822 4.13996 1 14 .06128 .486 1.027Table 7.19 Results of curve fitting for TM5 prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00405 .05699 1 14 .81478 .437 -.002 Logarithmic .00415 .05841 1 14 .81253 .579 -.062 Inverse .00398 .05591 1 14 .81651 .315 1.566 Quadratic .00468 .03058 2 13 .96995 .558 -.011 .000 Cubic .00501 .03275 2 13 .96786 .539 -.008 .000 .000 Compound .00001 .00009 1 14 .99273 .312 1.000 Power .00034 .00482 1 14 .94563 .390 -.065 S .00149 .02084 1 14 .88726 -1.289 3.507 Growth .00001 .00009 1 14 .99273 -1.164 .000 Exponential .00001 .00009 1 14 .99273 .312 .000 Logistic .00001 .00009 1 14 .99273 3.202 1.000Table 7.20 Results of curve fitting for TM7 prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00354 .04978 1 14 .82668 .423 -.006 Logarithmic .00102 .01428 1 14 .90658 .426 -.025 Inverse .00000 .00006 1 14 .99392 .377 -.012 Quadratic .02408 .16037 2 13 .85349 .005 .101 - .006 Cubic .01733 .11461 2 13 .89260 .188 .039 .000 .000 Compound .04761 .69979 1 14 .41691 .167 1.083 Power .07054 1.06256 1 14 .32011 .066 .761 S .09848 1.52937 1 14 .23655 -.276 - 6.607 Growth .04761 .69979 1 14 .41691 -1.792 .080 Exponential .04761 .69979 1 14 .41691 .167 .080 Logistic .04761 .69979 1 14 .41691 6.004 .923 87
    • Table 7.21 Results of curve fitting for NDVI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .21662 3.87131 1 14 .06925 1.179 -1.238 Logarithmic .22868 4.15081 1 14 .06098 .020 -.810 Inverse .24056 4.43459 1 14 .05375 -.440 .522 Quadratic .26541 2.34853 2 13 .13467 4.518 - 8.188 11.765 Cubic .26541 2.34853 2 13 .13467 4.518 - 8.188 .000 11.765 Compound .22283 4.01419 1 14 .06487 6.248 .010 Power .21432 3.81902 1 14 .07095 .089 -2.871 S .20603 3.63282 1 14 .07739 -3.917 1.770 Growth .22283 4.01419 1 14 .06487 1.832 -4.601 Exponential .22283 4.01419 1 14 .06487 6.248 -4.601 Logistic .22283 4.01419 1 14 .06487 .160 99.556Table 7.22 Results of curve fitting for SIPI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .27346 5.26954 1 14 .03767 .517 -.524 a Logarithmic . . . . . .000 .000 Inverse .05916 .88035 1 14 .36401 .371 -.005 Quadratic .29356 2.70112 2 13 .10447 .514 -.816 .729 Cubic .33630 2.02686 3 12 .16389 .439 - 5.290 - 1.273 6.793 Compound .20593 3.63065 1 14 .07747 .494 .189 a Power . . . . . .000 .000 S .02640 .37969 1 14 .54766 -1.166 -.011 Growth .20593 3.63065 1 14 .07747 -.705 - 1.665 Exponential .20593 3.63065 1 14 .07747 .494 - 1.665 Logistic .20593 3.63065 1 14 .07747 2.024 5.286Table 7.23 Results of curve fitting for SR prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .17948 3.06236 1 14 .10200 .695 -.064 Logarithmic .20197 3.54324 1 14 .08074 .911 -.341 Inverse .22576 4.08228 1 14 .06289 .010 1.698 Quadratic .20869 1.71426 2 13 .21839 1.238 -.286 .021 Cubic .20869 1.71426 2 13 .21839 1.238 -.286 .021 .000 Compound .26727 5.10675 1 14 .04030 1.320 .750 Power .24007 4.42279 1 14 .05403 2.677 - 1.363 S .21588 3.85436 1 14 .06980 -2.463 6.081 Growth .26727 5.10675 1 14 .04030 .277 -.288 Exponential .26727 5.10675 1 14 .04030 1.320 -.288 Logistic .26727 5.10675 1 14 .04030 .758 1.334 88
    • Table 7.24 Results of curve fitting for NDWI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .23845 4.38350 1 14 .05497 1.068 -1.590 Logarithmic .24683 4.58802 1 14 .05026 -.241 -.734 Inverse .25127 4.69823 1 14 .04792 -.394 .329 Quadratic .25835 2.26418 2 13 .14332 2.392 -7.448 6.345 Cubic .26024 2.28664 2 13 .14096 1.993 -4.668 .000 4.744 Compound .28924 5.69733 1 14 .03165 5.155 .002 Power .28379 5.54725 1 14 .03362 .028 -2.882 S .27307 5.25914 1 14 .03783 -4.090 1.256 Growth .28924 5.69733 1 14 .03165 1.640 -6.417 Exponential .28924 5.69733 1 14 .03165 5.155 -6.417 Logistic .28924 5.69733 1 14 .03165 .194 611.893Table 7.25 Results of curve fitting for GVI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .10848 1.70357 1 14 .21287 .628 -.057 Logarithmic .11883 1.88801 1 14 .19103 .789 -.283 Inverse .12721 2.04047 1 14 .17509 .061 1.318 Quadratic .12593 .93645 2 13 .41693 1.101 -.260 .020 Cubic .12593 .93645 2 13 .41693 1.101 -.260 .020 .000 Compound .24609 4.56993 1 14 .05066 1.269 .731 Power .22303 4.01868 1 14 .06474 2.511 - 1.419 S .19798 3.45588 1 14 .08417 -2.589 6.025 Growth .24609 4.56993 1 14 .05066 .239 -.313 Exponential .24609 4.56993 1 14 .05066 1.269 -.313 Logistic .24609 4.56993 1 14 .05066 .788 1.368Table 7.26 Results of curve fitting for CVI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .05383 .79657 1 14 .38720 .670 -.133 Logarithmic .05319 .78655 1 14 .39013 .597 -.282 Inverse .05052 .74490 1 14 .40264 .115 .563 Quadratic .05405 .37139 2 13 .69686 .756 -.213 .018 Cubic .05434 .37352 2 13 .69546 .762 -.197 .000 .004 Compound .02420 .34721 1 14 .56509 .651 .721 Power .02897 .41772 1 14 .52853 .573 -.763 S .03278 .47441 1 14 .50222 -1.923 1.662 Growth .02420 .34721 1 14 .56509 -.430 -.327 Exponential .02420 .34721 1 14 .56509 .651 -.327 Logistic .02420 .34721 1 14 .56509 1.537 1.386Table 7.27 Results of curve fitting for MIRI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00175 .02461 1 14 .87759 .403 -.008 Logarithmic .00038 .00535 1 14 .94275 .393 -.015 89
    • Inverse .00018 .00251 1 14 .96074 .386 -.036 Quadratic .00470 .03072 2 13 .96982 .270 .060 -.008 Cubic .38110 2.46311 3 12 .11258 -6.037 5.025 - .096 1.241 Compound .08010 1.21906 1 14 .28816 .632 .823 Power .07106 1.07098 1 14 .31827 .774 -.725 S .05386 .79697 1 14 .38709 -1.845 2.303 Growth .08010 1.21906 1 14 .28816 -.459 -.194 Exponential .08010 1.21906 1 14 .28816 .632 -.194 Logistic .08010 1.21906 1 14 .28816 1.583 1.215Table 7.28 Results of curve fitting for GEMI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .21297 3.78845 1 14 .07196 .572 .000 a Logarithmic . . . . . .000 .000 Inverse .22649 4.09943 1 14 .06241 .213 -391.895 Quadratic .21448 1.77473 2 13 .20823 .605 .000 .000 Cubic .36816 2.33067 3 12 .12592 1.325 .001 .000 .000 Compound .28855 5.67804 1 14 .03190 .730 1.000 a Power . . . . . .000 .000 S .16009 2.66840 1 14 .12464 -1.653 - 1206.901 Growth .28855 5.67804 1 14 .03190 -.314 .000 Exponential .28855 5.67804 1 14 .03190 .730 .000 Logistic .28855 5.67804 1 14 .03190 1.369 1.000Table 7.29 Results of curve fitting for EVI prediction of relative abundance of A. germinans. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .19564 3.40523 1 14 .08624 1.595 -.565 Logarithmic .20591 3.63015 1 14 .07749 1.328 -1.243 Inverse .21663 3.87140 1 14 .06925 -.892 2.721 Quadratic .31536 2.99406 2 13 .08521 16.502 - 3.239 14.497 Cubic .31536 2.99406 2 13 .08521 16.502 - 3.239 .000 14.497 Compound .18986 3.28097 1 14 .09159 25.702 .130 Power .18758 3.23248 1 14 .09378 8.812 -4.345 S .18541 3.18646 1 14 .09593 -5.449 9.220 Growth .18986 3.28097 1 14 .09159 3.247 -2.040 Exponential .18986 3.28097 1 14 .09159 25.702 -2.040 Logistic .18986 3.28097 1 14 .09159 .039 7.687Table 7.30 Results of curve fitting for TM4-TM2 prediction of relative abundance of A. germinans. Model Summary Parameter EstimatesEquation R Square F df1 df2 Sig. Constant b1 b2 b3Linear .21500 3.83431 1 14 .07045 .742 -.008Logarithmic .21252 3.77822 1 14 .07230 1.661 -.334Inverse .19883 3.47436 1 14 .08343 .082 13.195Quadratic .21512 1.78153 2 13 .20712 .771 -.009 .000Cubic .21512 1.78153 2 13 .20712 .771 -.009 .000 .000 90
    • Compound .23970 4.41386 1 14 .05424 1.303 .971Power .20347 3.57624 1 14 .07949 31.586 -1.198S .16649 2.79650 1 14 .11666 -2.136 44.230Growth .23970 4.41386 1 14 .05424 .264 -.029Exponential .23970 4.41386 1 14 .05424 1.303 -.029Logistic .23970 4.41386 1 14 .05424 .768 1.030Table 7.31 Results of curve fitting for ARVI prediction of relative abundance of A. germinans. Model Summary Parameter EstimatesEquation R Square F df1 df2 Sig. Constant b1 b2 b3Linear .17533 2.97657 1 14 .10647 -.068 .209Logarithmic .14914 2.45400 1 14 .13954 .079 .403Inverse .12168 1.93943 1 14 .18545 .736 -.738Quadratic .27325 2.44392 2 13 .12560 1.746 -1.537 .406Cubic .26523 2.34631 2 13 .13489 1.111 -.647 .000 .060Compound .11942 1.89855 1 14 .18987 .083 1.880Power .10051 1.56442 1 14 .23152 .129 1.212S .08014 1.21970 1 14 .28804 -.082 -2.194Growth .11942 1.89855 1 14 .18987 -2.493 .631Exponential .11942 1.89855 1 14 .18987 .083 .631Logistic .11942 1.89855 1 14 .18987 12.096 .532Table 7.32 Results of curve fitting for TM4-TM3 prediction of relative abundance of A. germinans. Model Summary Parameter EstimatesEquation R Square F df1 df2 Sig. Constant b1 b2 b3Linear .22237 4.00340 1 14 .06519 .761 -.007Logarithmic .22885 4.15472 1 14 .06087 1.843 -.369Inverse .23057 4.19531 1 14 .05977 .027 18.047Quadratic .22453 1.88200 2 13 .19150 .898 -.012 .000Cubic .22453 1.88200 2 13 .19150 .898 -.012 .000 .000Compound .25368 4.75878 1 14 .04669 1.426 .973Power .21947 3.93653 1 14 .06721 60.913 -1.322S .18796 3.24047 1 14 .09342 -2.307 59.689Growth .25368 4.75878 1 14 .04669 .355 -.027Exponential .25368 4.75878 1 14 .04669 1.426 -.027Logistic .25368 4.75878 1 14 .04669 .701 1.028Table 7.33 Results of curve fitting for TM3-TM2 prediction of relative abundance of A. germinans. Model Summary Parameter EstimatesEquation R Square F df1 df2 Sig. Constant b1 b2 b3Linear .10784 1.69219 1 14 .21432 .588 .031 aLogarithmic . . . . . .000 .000Inverse .09542 1.47677 1 14 .24438 .195 -1.095Quadratic .11849 .87369 2 13 .44054 .338 -.051 -.006Cubic .12646 .94095 2 13 .41529 .403 .000 .005 .001Compound .14744 2.42110 1 14 .14202 .783 1.144 aPower . . . . . .000 .000S .08727 1.33855 1 14 .26665 -1.786 -3.836Growth .14744 2.42110 1 14 .14202 -.244 .135Exponential .14744 2.42110 1 14 .14202 .783 .135Logistic .14744 2.42110 1 14 .14202 1.277 .874 91
    • Appendix D: ANOVA analysis in curve fitting for candidates for suitable SVI for predicting Simpson’s biodiversity (1-D). Analyses processed by SPSS Statistics 17.0. Selected SVI is highlighted.Table 7.34 Results of curve fitting for TM4 prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .03330 .48228 1 14 .49875 .385 .002 Logarithmic .04998 .73659 1 14 .40521 -.298 .202 Inverse .06743 1.01228 1 14 .33143 .783 - 15.179 Quadratic .21484 1.77862 2 13 .20759 -1.559 .061 .000 Cubic .21812 1.81327 2 13 .20203 -.957 .033 .000 .000 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.35 Results of curve fitting for TM5 prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .02950 .42553 1 14 .52476 .410 .005 Logarithmic .02495 .35825 1 14 .55904 .120 .133 Inverse .01982 .28305 1 14 .60305 .674 - 3.066 Quadratic .04118 .27917 2 13 .76083 .868 -.029 .001 Cubic .03923 .26541 2 13 .77095 .693 -.011 .000 .000 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.36 Results of curve fitting for TM7 prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .07985 1.21498 1 14 .28894 .359 .025 Logarithmic .08987 1.38242 1 14 .25930 .136 .206 Inverse .09387 1.45038 1 14 .24843 .763 - 1.545 Quadratic .11321 .82978 2 13 .45798 -.107 .144 - .007 Cubic .12087 .89369 2 13 .43285 .005 .093 .000 .000 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 92
    • a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.37 Results of curve fitting for NDVI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .09037 1.39095 1 14 .25790 .102 .702 Logarithmic .11189 1.76376 1 14 .20540 .775 .497 Inverse .13461 2.17770 1 14 .16216 1.092 -.342 Quadratic .46876 5.73542 2 13 .01638 -8.053 26.414 - 19.999 Cubic .47141 5.79695 2 13 .01586 -5.387 13.708 .000 - 10.402 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.38 Results of curve fitting for SIPI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .11787 1.87071 1 14 .19295 .476 .302 a Logarithmic . . . . . .000 .000 Inverse .13266 2.14129 1 14 .16547 .563 .006 Quadratic .34189 3.37678 2 13 .06591 .482 1.157 - 2.136 Cubic .34511 2.10787 3 12 .15263 .501 1.267 - 1.635 3.233 b Compound . . . . . .000 .000 a,,b Power . . . . . .000 .000 b S . . . . . .000 .000 b Growth . . . . . .000 .000 b Exponential . . . . . .000 .000 b Logistic . . . . . .000 .000Table 7.39 Results of curve fitting for SR prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .02743 .39482 1 14 .53989 .447 .022 Logarithmic .06179 .92207 1 14 .35322 .297 .166 Inverse .10700 1.67747 1 14 .21621 .778 - 1.025 Quadratic .49310 6.32293 2 13 .01208 -1.454 .797 - .074 Cubic .49310 6.32293 2 13 .01208 -1.454 .797 - .000 .074 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 93
    • a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.40 Results of curve fitting for NDWI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00590 .08313 1 14 .77733 .461 .219 Logarithmic .00943 .13321 1 14 .72059 .663 .126 Inverse .01358 .19271 1 14 .66737 .714 -.067 Quadratic .07099 .49671 2 13 .61962 -1.639 9.511 - 10.066 Cubic .07099 .49671 2 13 .61962 -1.639 9.511 - .000 10.066 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.41 Results of curve fitting for GVI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .00212 .02980 1 14 .86541 .526 .007 Logarithmic .01243 .17620 1 14 .68103 .440 .080 Inverse .03057 .44151 1 14 .51719 .692 - .567 Quadratic .30727 2.88313 2 13 .09198 -1.210 .751 - .075 Cubic .30278 2.82275 2 13 .09592 -.658 .393 .000 - .005 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.42 Results of curve fitting for CVI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .02083 .29783 1 14 .59383 .396 .073 Logarithmic .01867 .26640 1 14 .61382 .442 .147 Inverse .01567 .22284 1 14 .64416 .684 - .275 Quadratic .02241 .14902 2 13 .86301 .600 - .043 .118 Cubic .02241 .14902 2 13 .86301 .600 - .043 .000 .118 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 94
    • a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.43 Results of curve fitting for MIRI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .02287 .32769 1 14 .57610 .646 - .025 Logarithmic .02642 .37986 1 14 .54757 .688 - .106 Inverse .02916 .42057 1 14 .52715 .436 .406 Quadratic .02973 .19918 2 13 .82186 .824 - .011 .116 Cubic .03189 .13177 3 12 .93930 .405 .214 - .006 .071 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.44 Results of curve fitting for GEMI prediction of 1-D. Model Summary Parameter EstimatesEquation R Square F df1 df2 Sig. Constant b1 b2 b3Linear .01872 .26705 1 14 .61338 .506 .000 aLogarithmic . . . . . .000 .000Inverse .12483 1.99691 1 14 .17947 .662 255.186Quadratic .28075 2.53719 2 13 .11741 .126 .000 .000Cubic .28129 1.56556 3 12 .24892 .089 .000 .000 .000 bCompound . . . . . .000 .000 a,,bPower . . . . . .000 .000 bS . . . . . .000 .000 bGrowth . . . . . .000 .000 bExponential . . . . . .000 .000 bLogistic . . . . . .000 .000 95
    • Table 7.45 Results of curve fitting for EVI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .10295 1.60674 1 14 .22563 -.219 .360 Logarithmic .11861 1.88395 1 14 .19148 -.077 .827 Inverse .13540 2.19238 1 14 .16085 1.436 -1.887 Quadratic .63913 11.51211 2 13 .00133 -27.890 26.221 - 6.012 Cubic .63913 11.51211 2 13 .00133 -27.890 26.221 - .000 6.012 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.46 Results of curve fitting for TM4-TM2 prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .03745 .54470 1 14 .47268 .423 .003 Logarithmic .06019 .89658 1 14 .35976 -.043 .156 Inverse .07842 1.19137 1 14 .29348 .718 - 7.268 Quadratic .20440 1.66995 2 13 .22621 -.496 .044 .000 Cubic .21841 1.81640 2 13 .20154 -.250 .026 .000 .000 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.47 Results of curve fitting for ARVI prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .03990 .58185 1 14 .45825 .742 - .087 Logarithmic .03213 .46474 1 14 .50654 .677 - .164 Inverse .02467 .35415 1 14 .56127 .415 .292 Quadratic .08292 .58774 2 13 .56969 -.312 .928 - .236 Cubic .08054 .56936 2 13 .57938 .047 .417 .000 - .036 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000 96
    • Table 7.48 Results of curve fitting for TM4-TM3 prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .04084 .59615 1 14 .45289 .412 .003 Logarithmic .06955 1.04651 1 14 .32367 -.153 .178 Inverse .09952 1.54725 1 14 .23397 .758 - 10.400 Quadratic .27945 2.52088 2 13 .11880 -.854 .052 .000 Cubic .28210 2.55416 2 13 .11599 -.463 .028 .000 .000 a Compound . . . . . .000 .000 a Power . . . . . .000 .000 a S . . . . . .000 .000 a Growth . . . . . .000 .000 a Exponential . . . . . .000 .000 a Logistic . . . . . .000 .000Table 7.49 Results of curve fitting for TM3-TM2 prediction of 1-D. Model Summary Parameter Estimates Equation R Square F df1 df2 Sig. Constant b1 b2 b3 Linear .02933 .42309 1 14 .52593 .460 - .014 a Logarithmic . . . . . .000 .000 Inverse .06905 1.03841 1 14 .32548 .691 .817 Quadratic .23628 2.01103 2 13 .17340 -.505 - - .332 .024 Cubic .24997 2.16634 2 13 .15417 -.212 - .000 .001 .183 b Compound . . . . . .000 .000 a,,b Power . . . . . .000 .000 b S . . . . . .000 .000 b Growth . . . . . .000 .000 b Exponential . . . . . .000 .000 b Logistic . . . . . .000 .000 97